I .II". I.I ‘: .IIII: .A. f .1“... .“4' ‘I. I.~S.."I“CC I. MAI I.I I. u... LI. . I.I- ‘5} .21. P- I.. _.; “‘"1—|I‘.‘:.I.‘L:‘fi"( f3?! . I.I...III -1 1 I .‘2,I 1 I.-.I IE- :1...- I1“: :L\:.‘~ I. I C::.~I.I.:.~ .I. I. II.:. .I it... I.. . vhf-t I ‘2..: 3...“ II..I.I. .Z.I~J- L' .1.“ 1.2.3.11 :2. guys—cu: I.I.I..II. .I..I I. I I I.. . I I t.-._C..‘a‘.v_ slur). 4 .94; .I I.I.I I.I. .. I .-.u. .. I " I II, 1 I. .m....- I. II . I‘. -. .I I.I...I. . - I ..._ I M. .1: I... 1.41“..- $1.11....» I.‘_I.I.-I. 11-3.... .II.-.I.-.I._.I. I.I-.u .~I I.-I.-- I.I.-IpI. a. .--.I. “-11-". I..”‘LI...I.II I. .I.. I... 1;...I.II In...“ .44..“ . ‘ ‘ .. .I.I. RI. -.I.LC.I I.I I.. I...II L ‘ “"1““ Harri... 1:; _ . I.I L'rk‘uIl‘ . I .. flung... .I. . A I 1 d ... . .I» "_ .I. I. .‘ - I.I“...I‘I... . 129:. \ .1. .. . “4;“...- I.I..I. II I Li..... I. I I.Iw. ..I,I.I.I.-I.I_I I.I. ..,;. .I. 4“. -I. .I I.I. I» ‘ «3-. ‘ ‘ I. I. I.. ‘ ~ _ .I... I- _..;..,I ~ I I ‘.. ....I.. I .I “I.I: I. II. ‘ 'I . ..I ...I I..-1-1a}; . I -...... .-. . I.I .. - “.515... I_:,.I.. .‘le'LI III... .I. ”I.....Im... 1-....- .1. I -. - ..I 51...“. :IEAII... I: I . I: > . “‘ “3 .‘II: hr. " .I.: ,. . :3- I . I 51‘: .-.......I.I.-=I. . . . . . I.-. .I . . - ~ .1- .7 I I.) ....I..I::;. , _ . ‘ “ ‘ ‘ I. quIVII \. lfflv ‘ \IZI. I I-Ir. .I.I 12.1.: . .. ..I.....I..I .. .. . . I U I.....“ I. I I» I11...._...I:I.I. I.I _II .1, $1.3:- I.. I. . .. .Im ., .-.. .I. 2.0-. -..II .- .I».-':-..II I I, “I.I. .. .II.:,.._.,.I. . I.I.I .5-.. ,.». a. ”'2: .I.. I .I- n-.4~I\.‘.kv I II. . . . . ("111/ 'I. .zz‘L‘J‘fiu. 1. 2:12,! .r. . . I-IIM I}, 4 I'I'I 'IIIIII‘IAJ 2,; , 32931 01527 2879 {-0 m-‘MM--l-~h —~ 7“ “D —"_o'.-UMW‘—Ir.-‘ I < 'c ‘ A ~" 1 1 a ' . '= t .._ ‘ '. g 1 I g 1 Q t" I" 7 " 1 {'4' - ‘l' f, { H-JD' r‘ ' i > 15,, I it "v.- i! i - ' ' ‘N ‘ ’ - é } l E Wr"r"cwmmav=1; i I . ' . a _ . I 4 , ' v I . ‘ I '....---.,,. - .- .-4 _ , _ .__ ._ _._._,..__! PLACE IN RETURN BOX to remove this checkout from your record. TO AVOID FINES return on or before date due. DATE DUE DATE DUE DATE DUE MSU Is An Affirmative Action/Equal Opportunity Institution MWMS-p. _—i—Wi LIBRARY Michigan State ' 7 University This is to certify that the thesis entitled THE EFFECT OF JUDICIAL CHARACTERISTICS ON THE SENTENCING OF FELONY OFFENDERS presented by . Ann Marie Kazyaka has been accepted towards fulfillment of the requirements for Master oLScienchegree in _Q_n_rimi a; Justice / . V Major professor Dale W 0-7639 MSU is an Ajfirmative Action/Equal Opportunity Institution THE EFFECT OF JUDICIAL CHARACTERISTICS ON THE SEN'I‘ENCING OF FEIDNY OFFENDERS BY Ann Marie Kazyaka A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE SCHCXJL OF CRIMINAL JUSTICE 1987 ABSTRACT THE EFFECT OF JUDICIAL CHARACTERISTICS ON THE SENI'ENCDQG OF FEIONY OFT'ENDERS By Ann Marie Kazyaka This study is an examination of the effects of judicial characteris— tics on the sentencing of 1,045 felony offenders in Detroit Recorder's Court in 1977. In particular, the relationship between judicial race and sentencing was investigated. Multivariate logit and OLS regression techniques were enployed to examine the decision to incarcerate, the length of minimum sentence given imprisonment and the proportion of the statutory maximum imposed. The analysis revealed that, with few exceptions, judicial characteristics did not affect the patterns of sentencing produced by the judges in the sample. A detailed analysis of judicial race, however, indicated that the factors relied upon in reaching sentencing decisions differed for black and white judges even though the outcomes produced were similar. No evidence of discrimination against defendants due to race was discovered. ACKNO’VIELIMEN’I’S I would like to thank my Committee for their patience and guidance throughout this long process. In particular, Dr. Thu Bynum was instrumental in helping me to recognize my limits. I am also grateful to Dr. Marvin Zalman for allowing me to reanalyze the data from the Michigan Felony Sentencing Project and to Garrett Peaslee for giving me access to demographic infonnation on the judges serving Detroit Recorder's Court in 1977. II. III. TABLE OF CONTENTS LIST OF TABLES STATEMENT OF THE PROBLEM A. Race of the Defendant B. JUdicial Characteristics C. Judicial Race D. The Present Study A REVIEW OF THE LITERATURE A..A Theoretical Framework B. Racial Discrimination in Sentencing C. Individual Characteristics and Judicial Behavior 1. variation Due to Individual Judges 2. variation Due to Judicial Tenure 3.‘Variation Due to Age 4. variation Due to Gender 5. variation Due to Race METHODOLOGY A. Data Sources B. Dependent variables C. Independent variables D. Research Questions ii iv 14 16 16 ,122 36 38 41 44 46 49 54 54 57 63 70 IV. FINDINGS A. The Choice on Type of Sentence B. Choices on the Magnitude of the Sentence _ 1. Analysis of the Length of Minimum Sentence 2 . Analysis of the Proportion of the Statutory Maximum Imposed C. The Process of Imposing Sentence V. DISCUSS ION AND CONCLUSIONS m'I'ES REFERENCES 73 73 92 96 106 119 128 141 147 4B U1 6B 7A 8A LIST OF TABLES Descriptive statistics for the decision to incarcerate and independent variables in the model (N=l,045) Probability of being incarcerated given the race of the defendant Probability of being incarcerated given the race of the judge logit estimates of the effects of the offender's race on the decision to incarcerate, controlling for offense, offender and court case processing characteristics Model classification table Descriptive statistics for the judges included in the logit model Logit estimates of the effects of individual judges on the decision to incarcerate, controlling for offense, offender and court case processing characteristics Model classification table Logit estimates of the effects of offender and judicial race on the decision to incarcerate, controlling for offense, offender and court case processing characteristics Model classification table logit estimates of the effects of offender and judicial race on the decision to incarcerate, controlling for offense, offender, judicial and court case processing characteristics Model classification table Descriptive statistics for the racial interaction variables included in the logit model (N=l,045) iv 74 76 76 78 78 82 83 83 85 85 86 86 88 10A 10B 11A 12 13 14 15 16 17 18 19 20 21 logit estimates of the effects of racial interaction variables on the decision to incarcerate, controlling for offender, offense and court case processing characteristics Model classification table Logit estimates of the effects of racial interaction variables on the decision to incarcerate, controlling for judicial, offender, offense and court case processing characteristics Model classification table Descriptive statistics for independent variables included in the Ordinary Least Squares analysis (N=592) Zero-order correlations between race variables and the continuous versions of the dependent variable OLS estimates of the effects of offender race on the minimum sentence in months, controlling for offender, offense and court case processing characteristics Descriptive statistics for the judges included in the model OLS estimates of the effects of individual judges on minimum sentence in months, controlling for offense, offender, and court case processing characteristics OLS estimates of the effects of offender and judicial race on minimum sentence in months, controlling for offense, offender and court case processing characteristics OLS estimates of the effects of offender and judicial race on the nunimum sentence, controlling for offender, offense, judicial and court case processing characteristics Descriptive statistics for the racial interaction variables included in the OLS models (N=592) OLS estimates of the effects of racial interactions on minimum sentence, controlling for offender, offense and court case processing characteristics OLS esthnates of the effects of racial interactions on minimum sentence, controlling for judicial, offense, offender and court case processing characteristics 90 90 91 91 93 95 97 99 101 102 104 105 107 108 22 23 24 25 26 27 28 29 30 OLS estimates of the effect of offender race on the proportion of the legal maximum imposed, controlling for offense, offender and court case processing characteristics OLS estimates of the effects of individual judges on the legal maximum imposed, controlling for offense, offender and court case processing characteristics OLS estimates of the effects of offender and judicial race on the proportion of the legal maxumun imposed, controlling for offense, offender and court case processing characteristics OLS estimates of the effects of offender and judicial race on the proportion of the legal maxumon unposed, controlling for offense, offender, judicial and court case processing characteristics OLS estimates of the effects of racial interaction variables on the proportion of the legal maxhnun hnposed, controlling for offense, offender and court case processing characteristics OLS estunates of the effect of the racial interaction variables on the proportion of the legal maxumrn imposed, controlling for judicial, offense, offender, and court case processing characteristics CImparisons of the factors affecting the decision to incarcerate for black and white judges (logit) Comparisons of the factors affecting the length of minumum sentence for black and white judges (OLS) Comparisons of the factors affecting the proportion of the statutory maximum imposed for black and white judges vi 110 113 114 115 117 118 121 123 125 CHAPTER I STATEMENT OF THE PROBLEM Justice is supposed to be blind. It treats all individuals alike--rich or poor, black or white, male or female. (Crockett, 1971: 339) More than any other actor in the criminal justice 1 system, the judge represents the ideal of justice. He is the one individual who is expected to refrain from taking sides in the case. Rather, he acts as the arbiter of justice (at least theoretically), mediating between the prosecution and defense. This perception of the judge as an impartial decisionmaker is so ingrained in the minds of Americans that the role of the judge "is the least questioned and most respected of all the participants" in the criminal justice process (Lefcourt, 1974: 259). According to Beccaria, the proper role of the judge is "merely to ascertain the fact" (1963: 21). This task, however, may not be as easy as it may seem at first. Frank (1949) observed that the facts presented to courts are primarily in the form of testimony given by witnesses. This testimony is extremely prone to error. Honest witnesses may give inaccurate testimony if they (1) erroneously observe the event at the time it occurred, (2) have a faulty recollection of an event accurately observed, or (3) unconsciously misstate the recollection of the event for any 1 number of reasons, including a bias against one of the parties that inadvertantly colors the testimony. Moreover, some witnesses are "downright liars" who perjure themselves on the stand. The assumption that courts will somehow be able to extract the truth from such a tangled web of truth, error and lies is dubious, especially given the fact that judges are themselves fallible. Judges may have faulty memories, make inaccurate observations or be affected by prejudice, thus making it difficult for them to "get at the facts". Facts, therefore, must pass through two imperfect lenses, first the witnesses and second the factfinders. Therefore, Frank concluded that "[a] trial court's findings of fact is, then, at best, its belief or opinion about someone else's belief or opinion" (Goldman & Sarat, 1978: 114), a belief that is shot through with subjectivity. Even assuming, arguendo, that judges can determine the facts, in many cases the facts are not clear cut. Thus, the decision could go either way--for or against the defendant. Judges recognize that in indeterminate cases, their decisions may differ from those of colleagues faced with similar situations. They may differ regarding the relevance of certain facts or the interpretation of certain legal rules. Most judges, however, present their decisions "as the inevitable result of logical reasoning" (Van Koppen & Ten Kate, 1984: 226). However, as Gaylin (1974) notes, although we expect judges to be "the detached descendents of Solomon", we must remember that they are human, with the same needs and motivations as the rest of society. Conferring the title of judge on an individual does not enlarge the person in any way. As Justice Black pointed out in his dissenting opinion in §£§§g y. g.§., "Like all the rest of mankind, they [judges] may be affected from time to time by pride and passion, by pettiness and bruised feelings, by improper understanding or by excessive zeal." For these reasons, some social scientists argue that judicial decisions should be treated as subjective choices for particular solutions to particular conflicts rather than the predetermined results of legal reasoning (Van Koppen & Ten Kate, 1984). Given the subjective nature of many judicial decisions and the very human nature of the individuals making them, these decisions need to be examined in order to determine what factors play a role in shaping them. In addition to the determination of guilt, judges make numerous other decisions; e.g., setting bail, ruling on motions, and sentencing convicted defendants. The sentencing decision is one in which subjective elements could easily come into play, since the judge is not determining a question of fact at this point. Rather, the sentencing judge attempts to impose an appropriate sanction given the characteristics of the offense and offender. Facts that have already been determined, i.e., mitigating or aggravating circumstances, provide the information that the judge employs in making the decision. The outcome, however, is shaped by the penal philosophy embraced by the judge (Hogarth, 1071). That is, a judge who believes in rehabilitation will sentence differently than one who emphasizes deterrence because these philosophies suggest different sanctions as appropriate. To a certain extent, the approach taken by the judge is determined by the institutional setting in which she works. Judges in states with indeterminate sentencing laws will, by definition, be working within a rehabilitative framework, while those faced with determinate or mandatory statutes will focus on the retributive or deterrent aspects of the sanction. However, there is much leeway in this decision for personal beliefs about the appropriate sanction to contribute to the dispostion of a case. There is much evidence suggesting that such beliefs do indeed influence decisionmaking (Gibson, 1983). Thus, the sentencing decision is one in which values and beliefs, as well as facts, play important roles in determining the outcome. In addition, judges have wide latitude in setting sentences. This is particularly true in jurisdictions which provide for indeterminate sentencing, where only a minimum and maximum sentence is set by statute. The judge, considering the specific characteristics of the individual case, selects a sentence covering a range of years that falls in between the statutory limits. The wide discretion enjoyed by judges in these jurisdictions inescapably leads to disparate sentences in factually similar cases. The issue for many critics of judicial discretion is whether the differences in case dispositions are based on legally relevant factors or if extra-legal factors, e.g., race or sex, influence the outcome. A. Race of the Defendant Since the recognition that extra-legal factors may influence the sentencing decision, a plethora of studies examining judicial behavior have appeared. The primary focus of this research has been on the issue of discrimination against defendants, particularly racial discrimination. (See Hagan & Bumiller, 1983, for a review of many of these studies.) Early studies addressing this issue found fairly consistent evidence of discrimination, however, this research failed to control for variables other than race which may legitimately affect sentencing, e.g., prior record and offense seriousness. If such variables are correlated in some way with the race of the defendant as well as the sentencing outcome, the finding of racial effects may be spurious. More recent studies which have controlled for these variables, as well as others, have had mixed results regarding the influence of race on sentencing outcomes. The question of racial discrimination in our courts, then, is far from settled. ‘L B. Judicial Characteristics If we are to understand the decisionmaking process, we must do more than examine case characteristics and impute the cause of disparate outcomes to the personality, attitudes or social background of the judges involved (Hogarth, 1971). Human beings are incredibly complex creatures; many factors shape behavior. Before we can conclude that case outcomes are due to any particular aspect of human nature, such as personality, research must rule out other possibilities, like social background or judicial socialization. Thus, research examining the effects of various individual characteristics on judicial decisions is required. A considerable amount of research has examined the effects of such variables as personality characteristics (Ziller et al., 1977; Gibson, 1981), social background (Cook, 1980; Goldman, 1979; Schmidhauser, 1960, 1979), values (Schubert, 1974; Atkins, 1974; Nagel, 1969), and the socialization of judges (Tate, 1981; Alpert, 1981) on judicial behavior. Much of this research has found significant relationships between individual characteristics and judicial decisions. However, the majority of these states have focused on appellate court decisions despite the fact that most cases never reach the level of appeal. Moreover, it could reasonably be argued that decisionmaking at the trial court level is qualitatively different from the decisionmaking occurring in the appellate court. Trial court judges decide issues of fact, whereas appellate judges examine legal issues, e.g., the proper interpretation and application of specific legal procedures. In addition, trial court judges hear and see all of the evidence first hand, getting the full impact of intonation, phrasing, etc. Appellate court judges, on the other hand, must rely on transcripts and other documents as the basis of their decisions, and are, therefore, more removed from the qualitative elements of the case. Perhaps, the individual characteristics of trial court judges play a more active role in decisions that are closer and more susceptible to the emotion found in the criminal courtroom. There is, then, a need for more research examining the effects of judicial characteristics on the decisionmaking of trial court judges, especially since their decisions directly the affect the liberty of convicted criminal defendants. C. Judicial Race There is also a need for research that examines the effect of judicial race on sentencing decisions. At present, studies examining the effect of this variable on case disposition are practically nonexistent (Uhlman, 1979). This absence of research is primarily due to the fact that the American bench is almost totally white. Only recently has the number of black judges increased sufficiently for interested social scientists to find adequate samples. As black representation on the bench improves, investigations of the effect of judicial race on decisionmaking will become more important. A number of reasons exist to believe that judges with diverse ethnic backgrounds might respond to similar situations in different ways. Nagel (1962) argued that ethnic affiliations correspond to different value orientations which, in turn, manifest themselves in different decisional tendencies. Although he could not examine racial differences in judicial decisions due to limitations in his sample, he hypothesized that black judges would make more liberal decisions than their white counterparts. As indirect evidence of this point, he noted that black voters were more likely to support programs that protect the interests of the underprivileged than white voters. In addition, the study found that national ancestry is related to the degree of liberalism in judicial decisions. Given that race may be a more potent, and certainly more direct, influence on the development of values than the national origin of one's ancestors, there is further reason to believe that black and white judges may reach different decisions. It has also been claimed that decisional tendencies are determined well before judges ever rise to the bench. One's environment while growing up shapes decisions and behavior throughout adulthood (Glueck & Glueck, 1964). If the environment which surrounds black children as they grow up differs in some significant way from that experienced by white children, black adults could be expected to react to situations in a different manner than whites. These differences would also be seen in judicial behavior. Judge George Crockett, Jr., acknowledged the role of personal history in shaping judicial decisions: A judge is product of his own experiences, of his own history, of the people from whom he came. So a black judge's exercise of discretion is not going to necessarily be the same as that of a white judge. But as long as it is reason, and the law as made by precedents established by white people, that discretion stands (1984: 202). The extent to which the discretion exercised by black judges differs from that employed by whites, is an important question in a society seeking to protect the ideal of impartial justice. A third reason to expect the decisionmaking of black and white judges to differ is the idea of "black consciousness". Most economically disadvantaged groups are politically inactive, especially in the more demanding forms of political participation, e.g., campaign work, involvement in community groups (Verba & Nye, 1972; Milbrath & Goel, 1977). The poor lack the resources, ability and motivation to participate in the political process. Yet, many studies have found that blacks are significantly more active in politics than other disadvantaged groups (Cf., Verba & Nye, 10 1972; Aberbach & Walker, 1972; Williams et al., 1973; Greenberg, 1974; Antunes & Gaitz, 1975; Steggert, 1975; Clemente & Sauer, 1975; London, 1975; Lamb, 1975; McPherson, 1977; Cohen & Kapis, 1978; Klobus-Edwards et al., 1978). Verba and Nye explain the greater participation of blacks in political activities in terms of black consciousness: If blacks participate more than one would expect of a group with a similar socioeconomic status (SES), the explanation may lie in the fact that they have, over time, developed an awareness of their own status as a deprived group, and this self-consciousness has led them to be more politically active than members of the society who have similar socioeconomic levels but do not share the group identity (1972: 157). Shingles (1981) cites this group identity as the factor which frees blacks from the idea that the responsibility for their oppressed condition lies with the individual, thus allowing them to transfer the responsibility to society and, consequently, the government. This realization has mobilized blacks to influence the policy-making process. The judicial decisions of blacks may be influenced by this sense of group identity. Tentative support for this hypothesis may be found in statements made by black judges regarding their role on the bench: This means that black judges should treat blacks with a special understanding, the understanding that springs from their own experience and from their experience with American history. And then, of course, you lean over backwards to be fair to whites (Wright, 1984: 349). 11 We organize as blacks and around our blackness because the melting pot ideology has been ineffective in producing meaningful progress in our heterogeneous society. We have come to know that respect for racial variations comes not through a disappearance of differences but by the retention of a sense of racial distinctiveness. Consequently, we believe that each ethnic group fundamentally must learn to accept itself, for equality is not similarity, but acknowledgement of valued differences. Further, we believe that this sociology of acceptance can be expressed in the law and developed as a means of delivering more individual freedom. The right to be black, therein established, will broaden the concept of justice and equality and can bring more freedom for blacks and whites alike (Howard, 1974: 382). Thus, a feeling of racial unity may lead black judges to behave differently than their white counterparts, especially when dealing with black defendants. Black judges may also see themselves as having a special role. Judge Crockett (1984), for instance, identifies three roles to be filled by black judges. First, black judges must teach the public about the inequities in the criminal justice system. Second, they must find the remedies for the "old evils that have plagued the poor and the underprivileged in our society" in the common law and apply them. Finally, she must also be a symbol of American Democracy. These special roles may lead black judges to make decisions that vary substantially from those made by whites faced with similar situations. Indeed, some black judges believe that their decisions should differ from decisions made by white judges: 12 If we do not have a radical criminology from the black judges, they can never justify their existence as judges. If we allow our judges, especially our black ones, to be called "your honor", simply because they don the black robe of their high office, then we deserve what we get as the end product. And, if we wish our black judges to be indistinguishable from the white ones, then there is no point in having black ones (Wright, 1984: 218). The American bench provides a tremendous opportunity for such a radical criminology to affect the treatment of disadvantaged individuals, to equalize the the power relationship between the state and defendant. Further, there is empirical evidence that black officials, e.g., mayors, city council members, school board members and legislators, do attempt to improve the condition of blacks by influencing policy decisions (Cf., Campbell & Feagin, 1975; Nelson & Meranto, 1977; Karnig & Welch, 1980; Welch & Karnig, 1979; Eisinger, 1982a and b; Dye & Renick, 1981; Feagin, 1970; and Meier & England, 1984). Judges, as black officials, may also attempt to influence the process through their decisions on the bench. It is also possible that these special roles perceived by black judges are more a function of membership in a minority group rather than directly connected to racial considerations. In this case, women and judges who as children came from lower class homes may also view themselves as instruments in improving the conditions of 13 disadvantaged groups in society. This is a question which needs to be examined empirically. It is therefore important for research to distinguish the effects of gender and childhood social class from the effects of race on judicial decisions. Unfortunately, information on the childhood of judges is difficult to obtain, so no studies have examined the issue of social class (Gibson, 1983). Moreover, the small number of women on the bench has, until recently, precluded analysis concerning gender. There is, then, a need for the inclusion of such variables in analysis of the effects of judicial race on case outcome. There are also reasons to believe that no differences between black and white judges will be found. First, despite the finding of statistically significant relationships between judicial characteristics and behavior in previous studies, these results were often substantively weak (Uhlman, 1979). Second, the strength of the processes of legal and judicial socialization may lessen or eliminate the impact of individual characteristics on decisions. Third, institutional and environmental pressures may result in conformity. For instance, judges in very large jurisdictions may be unable to make individualized decisions, which may be more susceptible to the influence of judicial characteristics, due to the high visibility of judges, lack of time and strained resources (Hagan & Bumiller, 1983). Finally, variation among individual judges 14 may lessen the impact of race. That is, some black judges may be very liberal while others are conservative. The net effect of race would be nullified if this variation is very great. D. The Present Study In light of the preceding discussion, the present study will examine the effects of judicial characteristics on the decision to sentence convicted felony defendants. The primary focus of the study will be on the effect of the race of the judge and defendant. In particular, the study will examine the direct and indirect effects of judicial race on decisionmaking. Researchers have noted that judicial characteristics may affect behavior in two possible ways (Hagan, 1974; Gibson, 1983). They may have a direct effect on the sentencing outcome, causing the sentence to vary with the characteristic in question—-in this case, race. There may even be an interaction effect between race of judge and race of defendant, resulting in sentences that depend on the racial combination of judge and defendant. Similarly, judicial characteristics may affect decisionmaking indirectly, even though case dispositions may appear to be independent of judicial characteristics. Black judges may assign different weights to factors that are considered in the sentencing decision. That is, the process of reaching the decision may differ even when the outcome obtained is similar. This study will address each of these 15 possibilities in an attempt to provide a more complete understanding of the sentencing process. CHAPTER II A REVIEW OF THE LITERATURE In order to appreciate the contribution of the present study to existing knowledge of the sentencing process, it is necessary to place the study within the context of a theoretical framework and previous research on the effects of offender race and judicial characteristics on case disposition. The first section of this chapter examines expectations of judicial behavior generated by the conflict and consensus approaches to criminological research. Substantive findings of past research on the effect of the race of the defendant will be discussed in the second section. Finally, results of studies examining the influence of the background characteristics of trial court judges will be presented. A: A Theoretical Framework Two contrasting views of human society have shaped the thought of social theorists since the time of Plato and Aristotle (Bernard, 1983). These views, conflict and consensus, present very different implications for an understanding of official reactions to criminal violations. Both approaches and their respective implications regarding l6 17 treatment of minority defendants will be examined in this section. According to the consensus view, laws reflect values that are universally shared by members of society. The function of the state, then, is to protect these values. To the extent that groups with conflicting interests exist within society, the state acts as a mediator between these groups while simultaneously representing the whole of society (Vold & Bernard, 1986). On the basis of the consensus approach, we would expect to find that defendants in like circumstances would be treated similarly independent of minority status. The state would react only to the need to protect those values shared by members of society. In addition, minority judges would be expected to behave in the same manner as majority judges for two reasons. First minority judges would share the same values as those in the dominant group and would, therefore, find illegal behavior equally repugnant and deserving of punishment. Second, since the laws would protect the needs of all members of society equally, minority judges would not feel a need to "even the odds" for disadvantaged offenders. Thus, we would expect to find no discrimination on the basis of either defendant or judicial characteristics. The conflict perspective, on the other hand, recognizes the extensive nature of pluralism in American society. Our 18 society consists of many diverse groups, each with its own set of values and interests. Thus, the assumption of universally shared values implicit in the consensus view is dubious. Moreover, only a small proportion of the members of society are able to influence legislation. Consequently, rather than representing the interests of all members of society, the state and law represent the interests of those groups that have sufficient power and resources to shape legislation and law enforcement. Since the interests of these powerful groups often conflict with those of minority groups, behavior of members of minority groups will be defined as criminal more frequently than the behavior of the powerful (Vold & Bernard, 1986). Based on the conflict view, we would expect minority defendants to be perceived as being in conflict with the interests of the dominant group, and therefore more likely to receive severe sentences than other defendants. Further, members of a minority group who are appointed to the bench might utilize the opportunity to advance the interests of minority defendants. Based on the conflict model, then, we would expect to find discrimination against less powerful defendants by majority judges and in their favor by minority judges. Blacks have long occupied a minority position in American society. Indeed, the U.S. has a legacy of discrimination against blacks. Historians cite numerous 19 examples of laws and articles of the Constitution that kept blacks in a subordinate position in society (Long, Long & Weston, 1979; Burns, 1973; Bell, 1973). These laws included those which allowed slavery; that barred blacks from certain places and occupations; and denied them freedom of the press, speech and assembly. Georges-Abeyies concluded that "racial discrimination and unequal treatment under the law was law in the United States (1984: 129). After the Civil War with the ratification of the Fourteenth Amendment and passage of federal laws prohibiting discriminatory treatment on the basis of race, the legal status of blacks began to improve. However, it was not until the civil rights movement of the 1960's that significant improvement in this area began to be felt (Uhlman, 1979). It was at that time that the courts began to take an active position opposing the unequal treatment of blacks in our society in such cases as Brown y. the Board 9: Education. Today, blacks have a stronger position in the law than ever before. Unfortunately, not all of the unequal treatment experienced by blacks was sanctioned by the law, i.e., due to their legal status, thus it was not completely eliminated by the anti-discrimination laws or more liberal interpretations of the Supreme Court. It is the presence of this illegal discrimination in the courts themselves that concerns the social scientists studying the effects of race on sentencing decisions. This interest is due to the basic premise that the law applies equally to all individuals in our society, 20 and therefore, treatment under the law should be similar for individuals in like circumstances. From a conflict perspective, then, blacks as a relatively powerless group in American society would be expected to receive more severe sentences than whites. Such an approach has been taken by some authors (Cf. Chambliss & Seidman, 1971; Quinney, 1975; and Turk, 1969). Further, black judges might be expected to sentence differently from their white counterparts, especially when dealing with black defendants. This would be particularly true if black judges do indeed see themselves as having a special role as an agent of social change (e.g., Crockett, 1983). It is important at this point to distinguish between the expected effects of race, social class and gender on case disposition. It has long been recognized that members of the lower economic classes do not have the power or resources to significantly advance or protect their interests (Greenberg, 1981). Thus, we might expect judges who came from lower class backgrounds prior to attaining the bench to utilize that opportunity to improve the conditions of lower class defendants. However, as noted in Chapter 1, considerable research has shown that blacks are more politically active than other disadvantaged groups. This may, at least in part, be due to a group consciousness on the part of blacks which has no equivalent in other groups. Thus, we would expect black judges to be more lenient with 21 defendants independent of social class. Like blacks, women have long held a subordinate position in American law. For many years, women were considered to be legally the property of their husbands or fathers, with little legal standing of their own. They have faced a long struggle to impove their legal status, and are still working towards equality under the law regardless of gender (Feinman, 1986; Epstein & Goode, 1971; Datesman & Scarpitti, 1980; Price & Sokoloff, 1982). Thus, we might initially expect female judges to sentence similarly to blacks. Most researchers looking at the gender issue, however, have focused on differences in the socialization of males and females (Kritzer & Uhlman, 1977). Women are socialized to be more nurturant, affectionate, and docile than men (Epstein, 1970). Therefore, we would expect women to sentence differently than male judges. In addition, if the socialization process is more influential on decisionmaking than the experience as a minority, we would expect to find that this effect differs from that found for race. To summarize, on the basis of the conflict model, the present study expected to find black defendants being punished more severely then similarly circumstanced whites. In addition, black judges were expected to be more lenient in their sentencing practices, especially when handling cases involving black defendants. Finally, female judges 22 were expected to sentence differently from males, but this effect would not necessarily be similar to that found for judicial race. B: Racial Discrimination in Sentencing One characteristic of the sentencing literature is the lack of a unified theory which guides the research (Hagan & Bumiller, 1983). While some theorists have attempted to link racial discrimination to the labelling and conflict perspectives (e.g., Chambliss and Seidman, 1971; Quinney, 1975; Turk, 1969), Hagan and Bumiller (1983) note that these viewpoints have little standing outside of sociology. Since many studies on sentencing come from disciplines other than sociology, these perspectives do not pertain to a good proportion of the literature. Hagan and Bumiller divide the research into studies which have taken the individual-processual approach or the structural-contextual approach. The individual—processual approach focuses on the characteristics of the defendant, offense and criminal justice processing as determinants of case disposition. It includes both the traditional sociological distinctions of legal, i.e., factors legitimately considered in the sentencing decision, and extra-legal factors, i.e., those that are outside the range of legitimate considerations. These studies, then, include examinations of such variables as prior record, race, sex, age, type of attorney, method of 23 disposition or seriousness of the offense. The structural-contextual approach, on the other hand, explains variation in sentencing through differences in the social context in which the decision takes place. This research looks at the ways in which such variables as the political context, the make up of courtroom workgroups, or the relative class positions of participants in the social structure on the sentencing decision. Recently, several thorough reviews of the sentencing literature have been published (Hagan, 1974; Hagan & Bumiller, 1983; Kleck, 1981; Klepper et al., 1983). These review have presented the substantive results of prior sentencing research and noted some of the methodological problems. Rather than duplicate their efforts, this section will briefly summarize the main points of the authors regarding the cumulative findings of sentencing studies in order to set the present study in the context of past research. In addition, several studies that are representative of advancements in the literature will be discussed individually. Volumes have been written on the question of racial 2 discrimination in sentencing. Hagan and Bumiller (1983) cite 51 studies addressing this issue published between 1934 3 and 1981 . The results of this research have been anything but conclusive, however. Fifty-five percent of these 24 studies found evidence of racial discrimination; 45% found no correlation between defendant race and case disposition. There is, then, considerable room for debate about what the studies have actually demonstrated regarding discrimination. Treating these studies as if they were cumulative may be misleading, however, because the empirical and qualitative meaning of race may have changed over time (Peterson & Hagan, 1984). That is, people may perceive and react to blacks differently than they did prior to the civil rights movement. Recognizing this possibility, Hagan and Bumiller dichotomized the studies into those analyzing data covering periods prior to 1969 or from 1969 and after. The 1969 date was arbitrarily chosen to reflect the time at which the authors felt the effects of the civil rights movement would most likely have been felt. Twenty-five studies utilized data from the first time period. Of these, 56% reported evidence of discrimination. This percentage decreases only slightly in the post-1968 data (53.8%), suggesting that the probability of discriminatory treatment has not changed significantly in spite of the civil rights movement, assuming that the 1969 date was an accurate measure of the effects of that movement. It is also possible that variables which could legitimately be expected to influence sentencing, e.g., type of offense and prior criminal record, are also related to the race of the defendant. If this is the case, the _i—f relationship between race and case disposition could be spurious. For instance, blacks may be more likely than whites to have long criminal records, which in turn lead to more severe sentences. It is, therefore, important to control for such variables when estimating the direct effect of race on sentencing (Garber et al., 1983). Thus, Hagan and Bumiller also divided the studies according to the presence or absence of controls for offense seriousness and prior criminal record. Seventy—five percent of the studies with no controls for offense and record found evidence of discrimination, while only 42% with such controls reported a racial effect. It appears, then, that at least some of the discrimination reported by the studies with no controls for legitimate factors may actually be due to spurious relationships. Hagan and Bumiller made an additional point regarding controls for legal factors in sentencing. Prior to 1969, only 27.3% of the studies employing such controls found evidence of discrimination. Fifty percent of the studies since that time, on the other hand, found a racial effect when controls for offense and record were utilized. The authors suggested that the reason for the increase in studies finding a racial effect lies in the tendency of researchers to more sensitively specify the structural contexts within which discrimination seems to persist. 26 A number of these studies reveal racial discrimination, for example: in rural but not in urban settings; among judges with culturally linked prejudicial attitudes; for crimes like rape and robbery that are inter-racial; among highly politicized crimes and settings; in cases in which probation officers offer presentence recommendations; and in conditions that mark the intersection of race and class positions in American society. In contrast, studies of the last decade that have not found discrimination have focused frequently on settings in which discrimination by race may be least likely to be expected, for example, in large urban jurisdictions and/or courts that handle large numbers of misdemeanor cases (Hagan & Bumiller, 1983: 31-32, citations omitted). The authors suggested that judges in large, urban, highly bureaucratized settings may simply be too constrained by the lack of resources and time, as well as the need for efficiency, to allow direct discrimination by race. Kleck (1981) took a somewhat more conservative approach to his review of the literature. Whereas Hagan and Bumiller accepted the conclusions reached by the researchers at face value, Kleck accepted only those conclusions that were supported by the data. In the small number of cases in which his assessment of the data differed from that of the authors, he recharacterized the findings to fit the evidence more accurately. In addition, he recognized that many times studies do not find unequivocal evidence one way or the other. That is, the analyses may suggest mixed results, so Kleck summarized the research in terms of whether they found consistent evidence of discrimination, had mixed results, or 27 found no relationship between race and case disposition. Finally, Kleck separated studies examining capital punishment from those investigating noncapital sentencing. This approach is reasonable because (1) capital sentencing is usually done by juries rather than judges, (2) capital cases may more directly involve an expression of social mores, and (3) decisions in these cases usually follow protracted litigation before the imposed sentence is final. Thus, capital cases may follow different patterns of disposition and discrimination than non-capital cases (Hagan, 1974). A total of 40 studies of discrimination in noncapital 4 . sentencing were reviewed . Of these, only 20% discovered cons1stent evidence of racial discrimination. Thirty percent found mixed results--one-third to one—half of their findings were in favor of the discrimination hypothesis. A full 50% found no relationship between race and case outcome. Based on these results, Kleck concluded that there is little evidence of racial discrimination in noncapital I» ‘ 7,1.“ sentencing. Moreover, the evidence becomes even weaker when only those studies that provide a control for the prior record of the defendant are considered. Twenty-three studies controlled for this variable. Only 9% of these found consistent evidence of discrimination. Thirty-five percent found mixed results, while 56% found no evidence of a racial effect. It appears, then, that while there is some 28 evidence of discrimination in sentencing, it is inconsistent. Furthermore, many studies that report a statistically significant relationship between race and sentencing fail to examine the magnitude of this relationship. Hagan notes the need to distinguish between statistical and substantive significance: A relationship is considered statistically significant when we have established, subject to an accepted risk of error, that there is a relationship between two variables. Separate from the issue of whether or not a relationship exists is the question of how strong the relationship is (1974: 361). Since tests of significance are influenced by sample size, it is possible to reach statistical significance when the relationship is very small. .Sentencing studies tend to utilize large samples, and are therefore prone to this problem. In his review of eight non-capital sentencing studies, Hagan found that the strongest reported relationship between defendant's race and case outcome improved the predictive accuracy of the model by only 8% (1974: 363). On the basis of these findings, Hagan concluded that knowledge of the defendant's race does not contribute substantially to an understanding of the sentencing decision. Although the reviews indicate that the evidence of 29 racial discrimination is weak and inconsistent, it is important to interpret theSe results with caution. Qualitative differences in the methods employed in sentencing studies make comparisons difficult. Statistical techniques in these studies range from simple bivariate analyses to sophisticated multivariate regression techniques, e.g., path analysis. Many social scientists argue that the inconsistent findings in the literature regarding race derive from deficiencies in the methods and statistics of prior research (Cohen & Kluegel, 1978; Green, 1961; Hagan, 1974; Wellford, 1975). In particular these authors cite the failure of past research to provide adequate controls for prior record and offense seriousness, and/or to use the more rigorous multivariate statistical techniques. In addition, while studies have examined the effects of numerous variables (e.g., weapon use, socioeconomic status, type of attorney, number of charges, and marital status, among others), many of these variables have not been consistently measured and considered in the literature (Hagan & Bumiller, 1983). It is therefore difficult to make any firm statements concerning the cumulative knowledge gained from research on discrimination. The remainder of this section examines three specific studies of sentencing. Each attempts to correct for problems noted by reviewers in prior sentencing research. 30 Specifically, sentencing research has been criticized for its failure to (1) utilize rigorous mulitvariate statistical techniques which allow for the simultaneous control of theoretically relevant variables, (2) control for legally relevant considerations in case disposition, (3) examine the effects of earlier decisions in the process at the sentencing stage, and (4) control for the political and social contexts within which such decisions are made. Although none of these studies are without methodological flaws, they represent a significant improvement over much research that has been conducted in this area. Unnever (1982) attempted to overcome three of the problems noted in past sentencing research: (1) the failure to use rigorous mulitvariate techniques which allow for the simultaneous control of other theoretically important variables, (2) the lack of stringent controls for seriousness of the offense and (3) the failure to consider the possibility of organizational discrimination. He utilized logistic regression to analyze data collected for a sample of adult male drug offenders convicted and sentenced between July 1 and December 31, 1971 in Miami, Florida (N=313). He modelled the effects of race and socioeconomic status on the decision to incarcerate, controlling for prior record, offense severity, type of attorney and the outcome of bail proceedings. The analysis consistently indicated that blacks and 31 hispanics are significantly more likelY.to receive a SghtEHEe of imprisonment than whites. In the model employing all of the control variables, the odds of blacks receiving a prison sentence were 2.583 times greater than the odds for whites. Hispanics were 2.344 times more likely to be imprisoned than whites (1982: 219). There was no evidence of direct economic discrimination. However, the data did indicate that those who could afford to hire a private attorney and make bail were less likely to be incarcerated, suggesting an indirect effect of socioecononmic status on case disposition. Unnever concluded that sentencing research can not ignore the effect of race on judicial decisions. Klepper and his colleagues (1983) have pointed out the possibility that discrimination at the sentencing stage may be masked by discriminatory decisions made by officials at earlier stages in the processing of criminal defendants. The criminal justice process is sequential in nature with discretionary decisions made by officials at each stage. Recognizing the potential effects of discrimination in earlier decisions on case disposition, Bernstein et al. (1977) examined the effects of several variables on three decisions: (1) the decision to fully prosecute a case or to terminate via dismissal, (2) the decision to adjudicate the defendant guilty or to adjourn in contemplation of 5 dismissal , and (3) sentence severity. Dummy variable 32 regression techniques were employed to analyze data collected on a sample of 1,213 males arraigned on a felony charge in a city in New York state between December, 1974 and March, 1985 whose case did not result in acquittal. Independent variables in the analysis included those that were related to the defendant's social attributes, those that might determine the reactor's expectations for and perceptions of certain offenders, those related to the organizational imperatives of the system, those related to the individuals doing the reacting, those summarizing the results of prior processes and those related to the offense. The analyses indicated that race did not affect either the decision to dismiss the case or the decision to adjourn in contemplation of dismissal. It did, however, contribute to the severity of sentences received by convicted defendants (b=.302, 1977: 750). Interestingly, and contrary to the expectations of the authors, the results suggested that whites received more severe sentences than non-white defendants. However, this relationship was weaker than those of other independent variables with sentence severity. Based on information obtained in interviews with judges, the authors suggested that judges assume the non-white subculture is more likely to accept deviant behavior from its members than the white culture. Judges allow for these cultural differences when making the 33 sentencing decision. Thus, crimes committed by non-whites seem less pernicious and, thereby, less deserving of severe treatment. Judicial expectations for white offenders, on the other hand, are higher, since there is little tolerance of criminal behavior in the white community. Thus, whites tend to be given more severe sentences. Perhaps the best research on the sentencing decision has been conducted by Hagan and his associates. Most sentencing studies hold constant the context in which sentencing occurs, causing the research to be static. Investigations of the effects of changes in context, such as the criminal consequences of a political movement, are impossible, leaving many important issues unresolved, especially from a conflict perspective. Hagan's work emphasizes the role of the social context in shaping case outcomes. One such study will be discussed here. Hagan and Bernstein (1979) examined data from a sample of 238 draft resisters convicted and sentenced during the period of 1963 to 1976. Multivariate regression techniques were utilized to examine the effects of several variables on the decision to incarcerate. Independent variables included defendant race, education, prior record, type of plea, the presence of a presentence report, the type of resistance (active or passive) engaged in by the defendant, the individual judge's predisposition to sentence in a particular way, the religion of the defendant and time period. The authors noted that 34 from 1963 to 1968 the government‘s policy regarding draft resisters was coercive in nature. From 1969 on, treatment of these offenders became more cooptive. They hypothesized that the use of imprisonment as a sanction for resisting the draft would, therefore, decrease in the second time period. The primary interest was in the ways in which changes in the political context influenced the effects of other independent variables on the sentencing decision. When the control for time period was not included in the model, the effect of race on the sanction imposed was not significantly different from zero. In the equation controlling for time period, however, blacks were found to be 10% more likely than whites to receive a sentence of imprisonment. Time period appears to suppress the effect of race; In an analysis examining the influence of the interaction of various independent variables and time period on sanction, the interaction between race, type of resistance and time period was found to be significant. Draft resisters who were both white and activists were more likely to be imprisoned during the period of cooptive control. Although the generalizability of these findings may be limited, this analysis highlights the need to examine the influence of social context on judicial behavior. One of the variables included in the analysis by Hagan and Bernstein measured the likelihood that an individual judge would assign a particular disposition. The variable was included in order to separate out the effect of individual judges. The authors noted a tendency in the literature to see case disposition as a "judicial wheel of fortune", with the outcome dependent on the judge assigned to the case. Thus, we hear that Judge X is a hanging judge, while Judge Y is famous for his leniency. Therefore, the analysis examined the question of whether the probability of imprisonment depended on whg did the sentencing. The results indicated that,.regardless of time period, the sentence imposed was significantly influenced by the individual judge. This finding points to a need for a more thorough examination of the effects of individual characteristics on judicial behavior. Gibson (1978) suggests that the inconsistency of the results of past studies of racial discrimination may be due in part to their choice of the institution as the unit of analysis. Research focusing on the behavior of the individual judge may produce different results. Hagan (1974) notes that future studies should investigate the importance of such variables as judicial cognitive styles, attitudes and perceptual patterns in shaping their decisions. 36 C: Individual Characteristics and Judicial Behavior Theoretical research-on decisionmaking has examined the influences of many different factors. Judicial decisions, like any other decisions, are affected by a complex set of variables (Gibson, 1983). Accordingly, studies of judicial decisionmaking have investigated the effects of attitudes (e.g., Schubert, 1974; Howard, 1981; Lamb, 1976; Goldman, 1975), role orientations (e.g., Becker, 1966; Gibson, 1977; Jaros & Mendelsohn, 1967; Ulmer, 1974), and environmental constraints (e.g., Atkins & Glick, 1976; Cook, 1979; Gibson, 1980; Kritzer, 1978, 1979; Eisenstein & Jacob, 1977). By and large, this research has indicated that judicial decisions are indeed influenced by these factors: In a nutshell, judges' decisions are a function of what they prefer to do, tempered by what they think they ought to do, but constrained by what they perceive is feasible to do. Roughly speaking, attitude theory pertains to what judges prefer to do, role theory to what they think they ought to do, and a host of group-institution theories to what is feasible to do (Gibson, 1983: 9). Background characteristics have been linked to the acquisition of attitudes and values (Gibson, 1983). That is, such characteristics are related to socialization processes which vary across different groups. These socialization processes result in the development of certain attitudes and values, which in turn influence behavior. For instance, 37 members of the Protestant religion are taught to follow the work ethic, while other faiths do not emphasize these values to the same degree. We would, then, expect Protestants and non-Protestants to differ in decisions where the values espoused in the work ethic come into play. It is not the case, however, that behavior, or even attitudes and values, follow directly from background characteristics. The effects of background variables depend on many factors, e.g., the social context of the decision (Gibson, 1983). Unfortunately, most studies addressing this question neglect to specify the processes linking background characteristics to attitudes, values or behavior (Grossman, 1967). Gibson (1983) has called for research clarifying the underlying processes at work in the acquisition of attitudes and role orientations. In the meantime, he suggests that such variables should be considered as imprecise indicators of larger theoretical concepts rather than a single class of causes of behavior. Grossman (1967) examined research on social backgrounds and judicial decisions. Although he focused specifically on research regarding appellate court judges, his more general recommendations apply to studies of trial court judges as well. He noted that future research should look to an understanding of the relationships of background characteristics to other factors which might be better predictors of judicial behavior, e.g., legally relevant 38 variables. He also suggested that analyses should be restricted to judges sitting an a single court over a relatively limited time period in order to control for differences in the environmental context of decisionmaking. One characteriStic of the research on judicial attributes is its focus on the behavior of appellate court judges. Very few studies have investigated the influence of individual characteristics on decisions made at the trial court level. As noted in Chapter One, there are reasons to believe that decisions made in the trial court differ significantly from those made by appellate judges. First, trial court judges determine issues of fact while appellate courts focus on legal issues. In addition, appellate court judges must rely on transcripts and other documents for their decisions, whereas trial court judges hear all of the evidence first hand. Consequently, it is possible that the effects of judicial characteristics on trial court decisions may also differ from those found in research concentrating on appellate judicial behavior. Therefore, the remainder of this section is restricted to a discussion of studies examining the characteristics of trial court judges as they pertain to the variables of interest in the present study. 1. Variation Due to Individual Judges Disparity in sentencing may reflect a consistent tendency of some judges to impose severe sentences and of 39 others to impose lenient ones. Several studies on the behavior of trial judges have examined this possiblity. In 1974, Partridge and Eldridge conducted an experiment designed to analyze disparity in the Second Circuit Court. Fifty judges were asked to impose sentence in 20 hypothetical cases for similar offenses on the basis of information contained in presentence reports. The sentences in each case were ranked according to severity relative to those imposed by other judges in the same case. The analysis indicated that 9n average most judges imposed sentences that were consistent with those handed down by other judges. However, individual judges were not consistent in their sentencing behavior across cases. Relative to one another, judges are sometimes lenient and sometimes severe. Only two judges could be said to be consistently severe, and only one was consistently lenient. The authors concluded that the inconsistency of judges in sentencing across cases suggests that individual approaches to this decision are more complex than is widely believed. Gibson (1978) was interested in differences in discriminatory treatment of offenders among individual judges. He analyzed data collected on 1,194 felony cases tried and sentenced between March, 1968 and October, 1970 in Atlanta, Georgia. An index of discrimination was created for each judge by subtracting the percentage of blacks receiving severe sentences from the percentage of whites 40 receiving severe sentences. This index was then compared to those created for other judges. The results revealed considerable variation between judges. At least three of the eleven judges considered were more severe when dealing with blacks than with whites, and one was significantly less so. By definition, the was significantly more severe in sentencing whites. Gibson concluded that blacks receive disproportionately severe treatment from some judges and are the beneficiaries of others. The overall treatment of whites, on the other hand, was more evenhanded. Frazier and Bock (1982) employed data on 309 criminal cases heard in Florida between June 1, 1972 to May 31, 1973. Initial analyses examined the means of characteristics of cases heard by the seven judges for whom sufficient data for a multivariate analysis were available. The results indicated that sentencing judges do not hear the same kinds of cases. Substantial differences were found for all of the traits considered with the exception of defendant age and type of attorney. Thus, some of the variation in sentencing patterns found in earlier studies may be due to differences in offender and offense characteristics. To investigate this possibility, Frazier and Bock estimated a series of regression equations. Each model regressed the decision to incarcerate on the individual judge controlling for other theoretically important variables. The analysis revealed no evidence of variation 41 in sentencing due to individual judges. Research considering this issue in the future, then, should control for offense and offender characteristics in the analysis. 2. Variation Due to Judicial Tenure Alpert and her associates (1979) presented four stages of socialization experienced by individuals making the transition from lawyer to judge. In Stage I, Professional Socialization, the individual receives formal legal training. Upon becoming a judge, Stage II, Initiation and Resolution, begins. During this period, the new judge makes adjustments to the specific demands of the judicial role. This period is followed by Stage III, Establishment, in which the individual settles into the role of judge and decides whether or not to continue his judicial career. The final stage (Stage IV, Commitment) is characterized by increased identification with the court and loyalty to the judicial role. The authors hypothesized that different stages of the socialization process would be characterized with different role orientations. A survey of federal judges revealed that 6 this was indeed the case . Newly appointed judges (Initiation and Resolution) are more responsive to the demands of the public in their orientation to the judicial role. Judges who have served longer on the bench (Establishment) tend to look to attorneys for cues, and 42 eventually view themselves as guardians of the law (Commitment). In a qualitative analysis of interviews with Florida trial judges, Alpert (1981) concluded that the socialization of state trial judges follows the same process 7 as federal court judges . Thus, we might expect that empirical studies including a measure of judicial tenure would find that more experienced judges are more conservative in their sentencing decisions. Further tentative support for this hypothesis is found in Hogarth's (1971) examination of the background characteristics of 71 Ontario magistrates. Although he did not directly consider the impact of judicial tenure on behavior, his findings suggest that more experienced judges do indeed View their role differently than those who are new to the bench. He found that the length of experience is associated with a more moderate and coherent penal philosophy. Judges are more likely to emphasize deterrence, rather than retribution or reformation, the longer they remain on the bench. In addition, judges with longer tenure are more likely to view the practices and principles of other magistrates as providing a general guide. Empirical analyses of the effects of judicial tenure on case disposition have had mixed results. Engle (1971) investigated the effects of judicial characteristics on sentence severity in a sample of 8,119 defendants convicted 43 in Philadelphia in 1964. He employed correlational techniques which allowed him to control for case characteristics. The analysis was conducted separately for each of 27 offenses. On the basis of the results of this analysis, Engle concluded that for some less serious offenses tenure plays an important role in determining the severity of the sentence imposed. Judges who have served longer terms on the bench tend to impose more severe sentences in such cases. However, in OLS analyses tenure explained only a small proportion of the variance in sentence severity relative to that explained by case characteristics, e.g., prior record. Moreover, Engle cautioned that the interaction between age and tenure might be so pronounced as to render the product-moment correlations invalid. In the study described earlier in this section, Frazier and Bock (1982) conducted further analyses in order to examine the effects of various judicial characteristics on the decision to incarcerate convicted offenders. Multiple regression equations which controlled for offense and offender characteristics were estimated. The results indicated that the number of years on the bench does not significantly affect the probability of imprisonment (b=.008, s.e.=.026; 1982: 267). The inconsistency of these findings may, in part, be due to the use of different dependent variables. That is, 44 judicial tenure may not influence the decision to incarcerate, while it may play a role in determining the severity of the sentence. Moreover, experience may be an important predictor of the sentencing decision in only certain types of crimes. Future research should investigate the effects of this variable on different aspects of the sentencing decision and for various offenses. 3. Variation Due to Age Similar to judicial tenure, the age of judges has been hypothesized to be related to conservatism. Older judges are expected to be more conservative, and consequently more punitive in their sentencing decisions. Empirical support for this hypothesis is limited, however. Hogarth (1971) found no statistically significant relationships between the ages of magistrates and their attitudes or penal philosophy. There were, however, significant relationships between age and certain beliefs which might lead to differential sentencing patterns. Older judges considered a larger number of factors to be essential to the determination of the sentence. They also tended to minimize sociological explanations of criminal behavior. In addition, rather than feeling pressure toward uniformity with other judges in their decision, they felt that the most that can be expected in the justice system is uniformity in the application of legal principles. There is reason, then, 45 to expect older judges to sentence differently from younger judges. Gibson (1978) reported a moderately strong bivariate correlatidn between judicial age and an index of discrimination. However, this relationship disappeared in a regression analysis controlling for other judicial attributes. Gibson concluded that the initial correlation was largely a function of the relationship between age and memberships in professional organizations. Thus, judges' propensity to discriminate does not vary with age. Engle (1971) also investigated the influence of judicial age. Contrary to his expectation that conservatism would increase with age, the data revealed that older judges tended to impose less severe sentences than younger judges. However, the amount of variation in the dependent variable explained by judicial age was small relative to that explained by other variables in the analysis. Overall, he concluded that judicial age does not contribute significantly to the sentencing decision. Frazier and Bock (1982) expected to find older judges to be more conservative and consequently more likely to incarcerate convicted offenders. Regression analyses including this variable yielded results which were inconsistent with this expectation. On the basis of these results, the authors concluded that there is no reason to 46 believe that the age of judges has any effect on the likelihood of imprisonment for offenders. 4. Variation Due to Gender Various reasons for possible variation in judicial behavior due to gender have been offered by social scientists. The most frequently espoused explanation focuses on differences in the socialization processes experienced by males and females. From early childhood, women are taught to be more docile, nurturant, affectionate and dependent than men (Epstein, 1970). As a result, there have been suggestions that women are more sensitive to violations of norms than men (e.g., Parsons & Bales, 1955), and will thus be more severe in sentencing offenders. A related explanation has to do with discrimination in the workplace. Women seeking a position on the bench must deal tremendous obstacles in order to achieve their goal. Consequently, once the position is attained, they feel that they must "work harder to be more efficient, to be more competent, and to avoid any mistakes" (Feinman, 1986). Thus, female judges may be more cautious in their decisionmaking, resulting in different sentencing patterns than those produced by male judges. On the basis of these two hypotheses alone, there is reason to examine the effects of judicial gender on sentencing decisions. Kritzer and Uhlman (1977) hypothesized that female 47 judges would be more threatened by challenges to norms and law than their male counterparts, especially in cases involving crimes of a sexual nature. They analyzed data from a sample of 23,560 cases docketed and disposed of in a large northeastern city (dubbed Metro City) between July 1, 1968 and June 30, 1974. Contingency table analysis was employed to cross-tabulate a variable representing the gender combination of judge and defendant with three judicial decisions--the verdict, the decision to incarcerate, and the average prison sentence. The analysis was conducted separately for eight offense categories. Chi-square analysis showed that there were statistically significant relationships between the gender variable and sentence imposed in all of the offense categories. However, none of the relationships reported were strong. Further analyses discovered that female judges were more likely to convict in manslaughter cases, less likely to convict in aggravated assault cases and tend to impose more severe sentences on convicted larceny defendants. The data also indicated that female judges were more severe when dealing with robbery and drug offenders. However, the authors noted that this finding was expected due to differences in the cases heard by male and female judges. Specifically, female judges were more likely to encounter defendants facing more counts and more serious charges. Overall, Kritzer and Uhlman concluded that "there 48 is no strong evidence to suggest that female judges are consistently harsher with criminal defendants than are their male counterparts" (1977: 83). Gruhl and his colleagues (1981) took a somewhat different view of gender differences. They noted that several studies have reported women to be more liberal than men on a variety of issues, including those related to criminal justice (Cf. Diamond, 1977; Erikson & Luttberg, 1973; Soule & McGrath, 1977). They therefore hypothesized that women would be more liberal in their sentencing decisions than men. In addition, they expected to find that female judges would be more severe with female offenders than male judges who are more likely to hold paternalistic attitudes toward women. The analysis was based on 32,529 cases decided between 1971 and September, 1979 in Metro 8 City . As in the Kritzer and Uhlman study, the dependent variables in the analysis were the verdict, the decision to incarcerate and sentence severity. Simple differences of means tests and multiple regression techniques were employed. Initial analyses suggested that female judges were slightly less likely to convict, but more likely to incarcerate and to impose more severe sentences. However, the differences were not substantively large. In addition, the results were not consistent when offenses were 49 considered separately. Female judges were more severe in certain types of cases and more liberal in others. In light of these findings, the authors concluded that female judges are not consistently more lenient than males. The data did reveal more consistent differences in the behavior of male and female judges toward male and female defendants. Female judges were more likely to imprison female offenders than male judges. Although the difference is not significant, female judges tend to impose slightly more severe sentences on female defendants. When dealing with males, female judges are less likely to convict and more likely to incarcerate upon conviction. These differences were smaller than those for female defendants, however. The authors concluded that one part of the impact of the presence of women on the bench has been the reduction of favored treatment toward female offenders. 5. Variation Due to Race Although several researchers have looked at the social backgrounds of black judges, very little work has been done which examines the way in which race may affect the decisions of judges (Gibson, 1983). In part, this is due to the small numbers of black judges on the bench, thus making adequate samples very difficult to find. To date, only two studies have investigated the decision-making of black judges. 50 Engle‘s (1971) analysis included an investigation of the effect of judge's race on sentence severity. The analysis revealed that non—white judges were more lenient than whites in some offenses and more severe in others. The inconsistency of the results could be due to the small number of non-white judges hearing cases in the sample (only 4 of 82). With so little variation, any differences found could be due to idiosyncracies of the cases in the sample rather than actual racial effects. Moreover, the amount of variance in the dependent variable explained by judicial race was very small. Overall, Engle concluded that race does not affect sentencing decisions. Uhlman (1979) examined the behavior of 16 black and 78 white judges in a large northeastern city that he calls "Metro City". All felony cases docketed and disposed of between July 1968 and June 1974 were included in the sample. He looked at two critical decision points, (1) the decision to convict during a bench trial and (2) the severity of sentence given conviction. In his analysis of the decision to convict, Uhlman found that black judges tend to convict more often in cases involving black defendants. This pattern was also found in cases decided by white judges. Comparatively, black defendants fared less well before white judges, but the differences in such decisions and those made by black judges are not large. He concluded that "(w)hile a disparity between black and white defendants 51 is noted, the difference is of the same magnitude and direction as the pattern established by white Metro City judges"(Uhlman, 1979:67). Similarly, an examination of the severity of sentences handed down to convicted felony defendants by black and white judges was conducted. This analysis revealed that black defendants received more severe sentences than white defendants regardless of the race of the judge. Overall, black judges tended to be slightly more severe in sentencing defendants than their white counterparts, however, the difference was not significant. Uhlman's study provides evidence that judges make decisions independently of their own racial heritage. However, the analysis suffers from several methodological flaws. First, since the data comes from only one city, the findings may reflect idiosyncracies of that particular area, thus limiting the generalizability of the results. Uhlman recognizes this limitation and suggests replication in other areas . A second problem is the failure to control for characteristics of the trial judge other than race. If other characteristics are correlated in some way with the race of the judge, failure to control for their effects may distort true differences in the behavior of white and black judges. For instance, if the length of time served on the bench is related to sentencing outcomes and black judges tend to be relative newcomers, failure to control for this 52 variable would lead to inaccurate estimates of the effect of judicial race on sentencing decisions. In addition, one could argue that potential differences in judicial behavior may be due to the individual's more general experience as a minority, rather than racial considerations. If this were the case, we would expect women, members of the lower class and blacks to make decisions similar to each other and different from those made by white, upper/middle class males. In order to rule out this possibility, the decisionmaking of other minorities on the bench also needs to be examined, thus suggesting the inclusion of such variables as gender and class in the analysis. Finally, even if case outcomes are similar regardless of race, it is possible that the factors which influence these decisions differ for blacks and whites. Such a finding would be important in the understanding of how race may affect the behavior of officials, even though disparity in outcomes does not result. Therefore, analysis of this question should include a model with variables found to be predictive of sentencing outcomes in previous studies. In sum, there is considerable room for more research on the effects of judicial characteristics on the decisions made in trial courts. Prior research on this issue is scarce, leaving us with little information with which to work. That which is available is characterized by inconsistent results. There is therefore, little cumulative 53 knowledge in this area. Moreover, given the paucity of research on the political decisionmaking of blacks, the importance of the issue, and the methodological problems in both studies addressing the question of racial effects on decisionmaking, there is a clear need for further investigation. In particuluar the question of whether judicial race affects case disposition has yet to be answered. The present study is an attempt to clarify some of these issues. CHAPTER III METHODOLOGY A . Data sources This study is a secondary analysis of a portion of the data collected by Zalman and his associates for the Michigan Felony Sentencing Project (1977). The data analyzed here consist of a stratified random sample of approximately 25% of all felony cases in which the defendant was sentenced in 9 Detroit Recorder's Court during the calendar year 1977. The project utilized a disproportionate sampling scheme designed to ensure that all areas of the state and levels of offense seriousness were represented in the sample. This sampling procedure resulted in an initial sample of 1,322 cases from Detroit Recorder's Court. Data were collected primarily from the presentence investigation report associated with each case. The present analysis used information collected on the offense, the offender and the court processing of the case. Zalman and his colleagues collected statewide data on felony case processing. As Grossman (1967) suggested, comparisons across courts risk being distorted by variation in the political and social environments of different jurisdictions. Therefore, the present analysis was limited to one jurisdiction, Detroit. Detroit's Recorders Court was 54 55 selected as the site of the analysis for several reasons. First, the court serves the largest metropolitan jurisdiction in the state, which made it possible to obtain a sample of cases of adequate size and with sufficient variation on a number of theoretically relevant factors to allow simultaneous controls for various offense, offender and court case processing characteristics. Second, Recorder's Court is unique in that it had a relatively large concentration of female and/or black judges on the bench in 1977 (14 blacks, 4 females out of a total of 68). As noted previously, the number of blacks and females on the American bench is quite small, thus courts with variation in the race and gender of its judges are quite rare. This feature made analysis examining the influence of judicial race and gender possible. In addition, Detroit itself is unique in the racial composition of its government. The mayor of the city is black and has implemented policies designed to enhance black participation in the administration of city government. Within a power structure oriented toward the large black population of the city, black judges might feel more free to make decisions which differ from those of their white colleagues. Clearly, Detroit and the court which serves it have several unique characteristics which provided an opportunity to examine issues related to race unavailable in other parts of the state. Since the Zalman et al. data did not contain any 56 information about the characteristics of the sentencing judge, these data were combined with data obtained from the Michigan State Court Administrative Office. This office was able to supply the author with information on the race, sex, age and length of judicial service for the 68 judges who 10 sentenced defendants in the sample. These data, in conjunction with the Zalman et a1. data, allow the simultaneous consideration of the effects the characteristics of the offense, offender and judge on the sentencing decision. The present analysis focused on the behavior of the judge at the sentencing stage of criminal case processing. Of special interest were the effects of the race of the 11 judge and of the offender on (1) the decision to incarcerate, (2) the length of the minimum sentence given incarceration, and (3) the proportion of the statutory maximum sentence imposed on the defendant given incarceration. In order to simultaneously control for theoretically relevant variables other than race, multivariate regression techniques were utilized to estimate the model. 57 B. Dependent Variables Much of the sentencing literature is affected by concern over which aspect of the sentencing process should be examined. Sentencing decisions are multidimensional. That is, case dispositions are the result of a series of qualitatively different decisions, e.g., the decision whether or not to incarcerate, the severity of the sentence, and the proportion of the legal maximum to be imposed. Typically, sentencing research has taken one of two approaches--examination of the magnitude of a single type of sentence or collapse of the various sentencing options into a single, arbitrary scale of sentence severity. Both choices are problematic. Studies which focus on the magnitude of a single type of sentence face the risk of several types of measurement error. Blumstein et a1. (1983) have noted three. First, the addition of a specific period of time to a sentence may mean qualitatively different things depending on the statutory maximum sentence. That is, an increase of one year would be a much more severe increment in the case of a crime for which the statutory maximum is five years than it would in offenses for which the statutory maximum is twenty years. Second, many studies of this type tend to code all other sentencing options available to the judge as zero. 58 This measurement scheme results in biased measurements of the effects of the independent variables on case outcome. Finally, studies that attempt to overcome the second problem by excluding cases in which other types of sentences were imposed suffer from selection bias. That is, if racial discrimination occurred at an earlier point in the selection process, e.g., the charging decision, it may be obscured at later points where the decisions are made more evenhandedly (Klepper et al., 1983; Thomson & Zingraff, 1981). The use of a single, arbitrary scale of sentence severity is also problematic. These studies generally assign values to sentence outcomes, imposing a ranking of the severity of sentences without empirical substantiation. Such an arbitrary ranking raises questions about the relative severity of different sentencing choices, e.g., whether a sentence of five years probation is more or less serious than one year in prison. The use of this type of scale is also prone to statistical errors (Blumstein et al., 1983). First, the scale introduces errors in the outcome variable, which leads to imprecision in the estimates of the effects of the independent variable. Since the scale is arbitrary, it is difficult to interpret the coefficients obtained. At most, the effects can be interpreted to be increases in arbitrary units, which may have little practical or theoretical value. Moreover, the effects associated with the arbitrary scale may have no relevance to 59 any single aspect of the sentencing decision. Thus, the researcher can not untangle the influences of independent variables on any specific type of sentence. In addition, such a scale assumes that the factors shaping the sentencing decision are the same and have the same weights, regardless of the type of sentence imposed. This may not be an accurate representation of the decisionmaking process. For instance, factors related to the defendant's ties to the community, e.g., marital or employment status, may be more influential in the decision to incarcerate than in determining the length of sentence given imprisonment. Blumstein et al. have suggested a different approach to investigating the sentencing decision. This approach partitions the sentencing decision into two parts: (1) a choice between different types of sentences and (2) a choice on the magnitude of the selected type. They further suggest the use of statistical techniques that allow for the nonlinearities imposed by categorical dependent variables, e.g., PROBIT or LOGIT. Research examining both measures of case outcome would avoid many of the problems noted above. In particular, studies including choices among sentence types would minimize the effects of selection bias. Moreover, this approach would be more flexible than the arbitrary scale in allowing the examination of the effects of various factors on different types of sentence outcome. If comparison of the relative severity of different 6O sentences is desired, use of empirical techniques to determine the appropriate ranking of outcomes is suggested. The present study has taken the approach outlined by Blumstein and his colleagues one step further. The sentencing decision was partitioned into three parts: (1) a choice between types of sentences (incarceration or placement in the community), (2) a choice on the magnitude of the selected type (the length of sentence given imprisonment), and (3) a choice on the magnitude of the selected relative to the most severe sentence which might be legally imposed the proportion of the maximum sentence received. The dependent variable ”decision to incarcerate" was a dichotomous variable, coded 1 if the defendant was sentenced to time in prison and 0 if the defendant was placed in the community (e.g., jail, probation or fine). Since this version of the dependent variable is discrete, ordinary least squares (OLS) regression techniques are inappropriate for two reasons. First, the variance of the disturbance term depends upon the values of the independent variables, thereby violating the assumption of homoscedasticity required for OLS regression. Violation of this assumption does not bias the estimates, but the estimator will no longer be the best linear unbiased estimator (BLUE). In addition, the variance of the estimates would be large, making tests of significance difficult. Second, OLS requires the specification of a 61 linear relationship between the dependent and independent variables. With a dichotomous dependent variable, as in the present case, however, there is an upper and lower boundary on the possible values of the predicted outcome. This implies that, at the very least, the relationship will be non-linear at the boundaries. An S-shape functional form is thus more appropriate. Estimation of a linear probability model when the true relationship.is curvilinear introduces a systematic error into the model which precludes obtaining good estimates of the parameters of the distribution (Hanushek & Jackson, 1977). LOGIT and PROBIT models are designed for the estimation of models in which the dependent variable is dichotomous. In this type of analysis, the underlying probability function is logistic and the logistic estimator is a maximum likelihood one. The estimated coefficient represents the change in the log of the odds ratio (dependent variable) associated with a unit change in 12 the independent variable . The remaining versions of the dependent variable, "minimum sentence" and "proportion of the legal maximum sentence" were continuous variables, thereby making OLS regression techniques appropriate. In the models examining these variables, only those cases in which the defendant was sentenced to prison were considered. Defendants who received life sentences were also eliminated from the sample because of the arbitrary assignment of a value to the term 62 "life". Sentencing judges may have a very different conception of the meaning of a life sentence from the definition imposed by the researchers collecting the data. Moreover, for some offenses (e.g., lst degree murder) a life sentence is mandatory upon conviction, leaving no room for judicial discretion in case disposition. Once cases involving nonincarcerative sentences, life terms, or obvious 13 coding errors were excluded, the final sample used in the analysis of decisions regarding the magnitude of the sentence consisted of 592 cases. The state of Michigan relies on an indeterminate sentencing scheme, where the judge selects a sentence spanning a range of years constrained by a statutorily defined minimum and maximum. According to the statute, the judge may not impose a minimum sentence which is greater than two-thirds of the statutory maximum. For the present analysis, minimum sentence was defined as the minimum sentence imposed on the defendant in months. The unit of months was used in order to allow for the possibility of incarceration for a period of less than one year. The variable proportion of the legal maximum was included in this analysis in order to examine the possibility that judges may impose sentences that are close to the legal maximum on the basis of race. As Blumstein et al. have noted (1983), an increase in sentence of just one month may be more severe than an increase of one year if the legal 63 maximum in the first instance is a year and in the second twenty years. This variable, then, allowed a more standardized analysis of sentence severity. It was defined as the minimum sentence imposed on the defendant divided by the maximum sentence that was set by the legislature given 14 the offense for which the offender was convicted C. Independent Variables The independent variables of primary interest were the race of the judge and the race of the offender. Judicial race was defined as a dichotomous variable which was coded 1 if the defendant was black and 0 if he/she was white. Similarly race of the offender was coded 1 if the defendant 9 12 was black and 0 if white. In the regression equations which controlled for the effects of other theoretically relevant variables, five offender characteristics were included in the equation: (1) gender; (2) level of education, i.e., whether the offender graduated from high school; (3) marital status; (4) employment status; and (5) prior record. Several studies have hypothesized that judges will treat female offenders more leniently than male offenders--the chivalry hypothesis (e.g., Gruhl et al., 1981; Atkinson & Neuman, 1970). The 64 present study addressed this issue as well. Offender's gender was entered into the equation as a dichotomous variable, coded 1 for males and 0 for females. Level of education and marital status were included as controls because there is reason to believe that these variables are in some way linked to race. It is widely held that blacks are not as well educated as whites and that black families tend to be single parent households. If this is true, we need to rule out the possibility that low levels of education and the lack of a nuclear family influence the judge's decision to incarcerate. These variables were entered into the equation as dichotomous variables which were coded 1 if the characteristc was present and 0 otherwise. The categorical variable "employment status" indicated the degree to which the defendant was employed at the time of the offense. It was entered into the equation as a series of dummy variables, i.e., full-time, part-time, and unemployed, with offenders who were employed full-time in the suppressed category. The data did not include a direct measure of socioeconomic staus. However, these three variables served as indicators of the offender's social position, and have been utilized in the past as proxies for SES (e.g., Myers, 1987). One of the strongest and most consistent predictors of case disposition in the sentencing literature has been the past criminal history of the defendant (Hagan & Bumiller, 65 1983). Prior record was a continuous variable defined as the number of adult felony convictions prior to the present case. Welch et a1. (1984) examined the influence of alternative measures of this variable. Their findings suggested that the number of felony convictions is significantly related to both sentence severity and the decision to incarcerate, even when type of offense is controlled. In addition, the relationships between this measure of prior record and the outcome variables were among the strongest observed. Further, preliminary analyses with different measures of prior record based on the Zalman et a1. data indicated that this measure was the strongest predictor of the sentence imposed. If the "best" measure of prior record is the one which is strongly related to the dependent variables, this measure was the most appropriate one available in these data. In addition to offender characteristics, each regression equation controls for several variables related to the processing of the case. The first of these concerned whether the defendant was able to hire a private attorney. It has been argued that discrimination may be based on economic factors rather than directly related to race (Kleck, 1981; Unnever, 1982). That is, defendants who cannot afford to hire private attorneys are hypothesized to be at a disadvantage in the court system. Since blacks are more likely to be unable to afford an attorney than whites, 66 sentencing studies need to control for the possiblity of economic discrimination in order to rule out the possiblity of spurious relationships. In addition, defendants who attempt to present their own cases are ill equipped to do so, and are thus more likely to receive more severe sentences. The type of defense attorney representing the offender was therefore entered as a series of dummy variables, i.e., private attorney, defended by self and public defender, with cases in which a private attorney was hired serving as the category of comparison. Similarly, if a defendant cannot afford to obtain release on bail, he may be more likely to receive a severe sentence. Therefore, a measure of the custody status of the defendant was included in the model. If the defendant was in custody at the time of sentencing, custodial status was coded 1, 0 otherwise. The prosecutor's charging decision is one of the most important decisions in the criminal justice system. It determines the extent of the individual's contact with the system (Vorenberg, 1981). Several authors have argued that prosecutors control the sanction imposed on convicted defendants (Wilson, 1975; Horowitz, 1977; Levin, 1977). On the other hand, some research has reported that alterations in the charge have no effect on the sentencing decision. These studies have indicated that judges tend to sentence according to the actual behavior of the defendant (or the amount of harm inflicted on the victim) rather than the 67 official charge for which he/she was convicted (e.g., Hagan, 1977; Wilkins et al., 1976). In order to examine this question, a variable to measure whether the arrest charge was reduced prior to conviction was included in the equation. If the charge for which the defendant was arrested was reduced, the dichotomous charge reduction variable was coded 1, 0 otherwise. In addition, a continuous variable measuring the total number of charges against the defendant was also included. This variable is coded 0 if the defendant faced one charge, 1 if he/she faced two charges, and 2 for three or more charges. This variable is important for two reasons: (1) defendants facing several counts would be more likely to receive severe sentences than those who were charged with only a single offense; and (2) the prosecutor is the individual who decides how many charges will be pressed. Therefore, this variable also indicates the potential of the prosecutor to shape the sentence imposed on the defendant. A second aspect of the prosecutor's decision is his ability to plea bargain. Defendants may plead guilty after negotiating with the prosecutor for a lenient sentence or a reduction in the charges. Moreover, a defendant may also enter a plea of guilty without such an agreement in the hopes that his cooperation will be rewarded by the judge with leniency. In order to discover if a plea of guilty does indeed result in less severe sentence than would be 68 expected if the case went to trial, a variable reflecting the method of conviction was included in the analysis. This variable was coded 1 if the defendant took the case to trial and 0 if he/She entered a plea of guilty. In those cases which go trial and result in a conviction, it was expected that the defendant would receive a more severe sentence. A second variable which has consistently been a strong predictor of case outcome is offense seriousness (Hagan & Bumiller, 1983). Unnever (1982) suggested that analyses be conducted within types of offenses with additional controls for seriousness within categories. In order to control for differences in the handling of different types of offenses, two offense variables were included in the model. On the presumption that violent offenses are, by their nature, more serious than property crimes, a variable was created which identified the conviction offense as either a crime against a person (e.g., homicide, rape, or assault) or one against property (e.g., burglary or larceny). Cases involving drug offenses, which are neither property or personal crimes and which constitute a sizeable percentage of the 1,322 cases (21%), have been eliminated from the sample leaving a final sample size of 1,045 cases to be used in the analysis of the decision to incarcerate. Cases involving violent crimes were coded 1; property offenses were coded 0. Since the categories of this variable were so broad as to encompass a wide range of behaviors varying in seriousness, a second 69 variable was created to control for the seriousness of the specific offense. This variable was defined as the maximum sentence which could be legally imposed on the defendant. The more serious the offense, the longer this statutory maximum would be. One of the suggestions in the literature has been that racial discrimination may be masked by variation in the sentencing styles of individual judges (Cf Gibson, 1978). That is, one judge may sentence very severely when dealing with blacks while another may be extremely lenient, thereby cancelling out the racial effect. In order to examine the possiblity that variation across individual judges accounts for disparity in case disposition, a model was estimated which included eight dummy variables representing individual judges. Each judge represented in the equation sentenced at least five percent of the cases in the sample to be analyzed for the respective versions of the dependent variable (N=l,045 for the decision to incarcerate; N=592 for minimum sentence and proportion of the maximum imposed). The variables were coded 1 if the cases was handled by that particular judge, 0 otherwise. Judges who did not sentence at least 5% of the cases in the sample were placed in the category of comparison. In addition to the previous models, a model including several judicial characteristics was estimated. This model included the gender and age of the judge, as well as a 70 measure of the length of time on the bench. Gender of the judge was coded 1 for male judges and 0 for females. The age of the judge was calculated by subtracting the year of his/her birth from 1977. Similarly, the length of time on the bench was determined by subtracting the year in which the judge first served on the bench from 1977. Finally, as noted in Chapter One, black judges may be more lenient with black defendants (Cf. Wright, 1984; Howard, 1974). Therefore, four variables were created to examine the effects of the interaction between the race of judge and defendant on the sentencing decision. Each variable is a dummy variable coded 1 if the specific racial combination is present in a case and 0 if absent. Analysis including these variables allowed examination of the possibility that black judges are more lenient when dealing with black defendants compared to other combinations of race of judge and defendant. D. Research Questions Briefly, the present study conducted analyses designed to answer the following questions: 1. Do sentences imposed on felony defendants in Recorder's Court follow a pattern based on the race of the offender? 71 2. Do individual judges significantly influence sentencing patterns? 3. Do the background characteristics of judges (i.e., gender, age and tenure) affect sentencing decisions? In particular, does the race of the judge or the racial combination of judge and defendant lead to differential outcomes? 4. Does the process of reaching the sentencing decision differ by the race of the judge even when outcomes are similar? That is, are different weights assigned to the factors considered in the decision if the judge is black or white? In order to address these questions, a number of control issues had to be addressed. Specifically, it was important to include variables found to be significant predictors of case outcomes in previous studies, e.g., prior record , offense type and offense seriousness. In addition, it was necessary to examine indicators of economic discrimination, e.g., type of attorney, custodial status, SES. Moreover, the effects of earlier decisions on outcome were investigated, e.g., total number of charges, charge reduction in order to alleviate the effects of selection bias. Inclusion of such control variables was necessary to rule out the possiblity that any pattern of racial discrimination was in fact due to other factors, thus 72 providing more confidence in the results of the analysis. CHAPTER IV FINDINGS A. The Choice on Type of Sentence Table 1 presents descriptive statistics for the dependent variable (decision to incarcerate) and the offender, offense, case processing and judicial characteristics in the model. The data indicated that 83.3% of the convicted offenders in the sample were black, and 16 92.3% were male . Only a small proportion of the defendants had at least a high school education (17.8%). The majority were unemployed (63.1%) or only working part time (7%) at the time of the offense. Approximately half were married. The average defendant had been convicted as an adult for one prior felony. Slightly more than one-third of the defendants were in custody at the time of sentencing. The average offender faced one additional charge. Fifty percent were convicted on charges that were less serious than those for which they had been originally arrested. Further, only 16% of the cases were disposed of through a trial; in the vast majority the defendant entered a plea of guilty. This finding is not unusual in itself, but it does highlight the potential power of the prosecutor to shape the sentence imposed on the 73 74 Table 1: Descriptive statistics for the decision to incarcerate and independent variables in the nodal (N'= 1,045) Standard variable Pfirt up; than: [twistion Decision to Inoaroerate 0 1 .634 --.. Offender: Gender 0 1 .923 ---- Race 0 1 .833 ---- Education 0 1 .178 -_-- Employment Part-time 0 1 .070 --__ unemployed 0 1 .631 ——-- Marital Status 0 1 .478 —--— Prior Record 0 5 1.122 1.508 Custodial Status 0 1 .365 ---- INunber of Charges 0 2 .848 .656 Charge Reduction 0 1 .500 ---_ Type of Attorney: Self 0 1 .083 -..- PUblic Defender 0 1 .677 -—-- Method of Conviction O 1 .160 ---- Type of Offense 0 1 .517 ---- Seriousness 2 300 131.283 105.170 Judge: Gender 0 1 .902 __-- Race 0 l .347 ---- Age 35 75 50.554 10.291 Tenure 1 42 8.065 5.436 75 defendant. Very few defendants attempted to represent themselves in court (8.3%); most relied on the services of the public defender. Approximately half of the defendants were convicted of a violent felony (51.7%), i.e., an offense involving violence or the threat of violence. The average statutory maximum sentence for these cases was approximately 131 months. It appears that more serious offenses were somewhat overrepresented in the sample. This may account for the fact that a large percentage of defendants received sentences involving a term of imprisonment (63.4%). Usually, the percentage of defendants sentenced to prison by Recorder's Court judges is much lower. For instance, in 1985 approximately 35% of the convicted defendants were sentenced to incarceration in a state prison (Michigan Department of Corrections, 1986). Ninety percent of the offenders were sentenced by male judges. Nearly thirty-five percent of the cases were heard by blacks. The average case was disposed of by a judge in his early fifties with approximately eight years experience on the bench. In Table 2 the probability of being incarcerated given the race of the offender is reported. Nearly 65% of all black offenders received prison sentences as compared to 57% of all white offenders, a difference of 8%. This finding 76 Table 2: PrObability of being incarcerated.given.the race of the defendant Race of the Defendant White Black Decision No .434 (76/175) .352 (306/870) to Incarcerate Yes .566 (99/175) .648 (564/870) Table 3: Probability of being incarcerated given the race of the judge Race of the Defendant White Black Decision No .352 (240/682) .391 (142/363) Incarcerate Yes .648 (442/682) .609 (221/363) 77 suggests that black defendants were slightly more likely to receive sentences involving imprisonment than were whites (gamma=.172). A similar analysis examining the relationship of the judge's race and the decision to incarcerate is presented in Table 3. These data indicate that black judges were not significantly more likely to send defendants to prison. Both races incarcerated slightly less than two-thirds of the defendants who were brought before them (gamma=-.084). The simple analyses presented thus far have indicated that black offenders were somewhat mere likely to be sentenced to a period of incarceration. One cannot make any firm conclusions based on these analyses, however, because they failed to control for the effects of other important variables, e.g., prior record, seriousness of the offense or method of conviction. The effects of these variables may tend to inflate or suppress the true effect of racial characteristics on the decision to incarcerate. In order to eliminate the possibility that the racial effects observed were in fact due to other factors a series of logit analyses which simultaneously control for theoretically relevant variables other than race were estimated. Table 4A reports the estimate of the effect of offender's race on the decision to incarcerate controlling for various offense, offender and court case processing characteristics. As can be seen in the table, the results 78 Table 4A: Logit estimates of the effects of the offender’s race on the decision to incarcerate, controlling fer offense, offender and.court case processing characteristics variable ‘b gi§.e. Offender: Gender .278 .310 Race .148 .230 Education -.554‘ .232 Employment Part—time .538 .361 unemployed .451‘ .203 Marital Status .109 .181 Prior Record .536*‘* .074 Custodial Status 2.414‘** .262 Number of Charges .118 .154 Charge Reduction —.111 .203 Type of Attorney: Self .758‘ .371 PUblic Defender .435‘ .211 Method of Conviction .553 .290 Type of Offense .093 .223 Seriousness .010‘*‘ .001 Constant —2.752*‘* .483 Log of Likelihood Function -418.331 Table 48: Model classification table W gut In Out 286 96 Actual In 100 563 Percent Correctly Classified.by Model: 81.25% Percent Correctly Classified by Chance: 53.49% Percent Reduction in Error Relative to Chance: 60.18% * p < .05 " p < .01 *** p < .001 79 were largely consistent with those found in previous sentencing studies. The probability of incarceration was significantly increased if the defendant had a previous history of criminality (b=.536, p<.001) and if the defendant was in custody at the time of sentencing (b=2.414, p<.001). The seriousness of the offense significantly affected the probability of imprisonment (b=.010, p<.001). There is evidence that the type of attorney played an important role in determining the likelihood of being sentenced to prison. It appears that private attorneys were better able to keep their clients in the community (for those represented by public defenders b=.435, and for those defendants representing themselves b=.758; p<.05). Aside from prior record, only two offender characteristics significantly affected one's chances of being incarcerated, level of education and employment status. Those individuals who had graduated from high school were less likely to be imprisoned than those without a degree (b=-.554, p<.05). In addition, those defendants who were unemployed at the time of the offense had a higher probability of being sentenced to time in prison than those who were employed full-time (b=.451, p<.05). The finding which was most pertinent to the current analysis was the lack of a significant effect on the decision to incarcerate due to the race of the offender. While the coefficient was in the expected direction, with 80 blacks being more likely to be sentenced to prison, it was nonsignificant (b=.148, p >.10). This result suggests that the statistically significant, but weak, zero-order association between the offender's race and the decision to incarcerate reported in Table 2 was due to the suppressed effects of variables related to the offense and court case processing, the defendant's level of education, and his/her employment status. In order to estimate the goodness-of-fit of the models produced from logistic techniques a Model Classification 17 Table is provided at the bottom of each table reported. For this first equation, the model produced a 60% reduction in the error one would have made by chance. In addition, the model accurately predicted 85% of the cases in which incarceration actually occurred. These findings suggest that the fit of the model was a reasonably good one. It is possible that the failure of this model to find a significant race effect was due to variation in the sentencing styles among judges. That is, some judges may have been consistently more likely to imprison black defendants while others were consistently less likely to do so. If this were the case, the effects of different judges would cancel each other out, making it appear that racial considerations did not enter into the sentencing decision. To test this possiblity, a model controlling for the effects 81 of individual judges was estimated. Only those judges who imposed sentences in at least five percent of the cases were included in the equation in order to ensure that no empty cells would occur. Table 5 presents descriptive statistics for the six judges represented in the model. As the table indicates, no single judge was responsible for sentencing a large proportion of defendants. This is not surprising, given the number of judges serving Recorders Court during the period covered by the study (N=68). All six are male, and five are white. Table 6 reports the results of the logit analysis which included variables representing the six judges. As the table demonstrates, these results were similar to those found in the first model. However, the effects of two variables that were significant in the previous model only approach significance once the judges were added to the equation. It appears that at least part of the variance explained in the first model by the unemployment of the defendant and his/her reliance on a court appointed attorney was due to the effects of individual judges. Two judges significantly affected the likelihood of incarceration. Judge 12 was significantly less likely than the judges not included in the equation to sentence a defendant to prison (b=-1.100, p<.05), while Judge 127 was more likely to do so (b=.825, p<.05). However, the effects of these judges did not appear to substantially alter the 82 Table 5: Descriptive statistics for the judges included in the logit model Percentage Judge of Cases Rice Gender 38 5.7 W M 75 5. 1 W M 112 5.6 W M 127 6.6 W M 130 6.4 W M 143 6.0 B M 83 Table 6A: Logit estimates of the effects of individual judges on the decision to incarcerate, controlling fer offense, offender and court case processing characteristics variable ‘b g.e, Offender: Gender .346 .322 Race .144 .262 Education -.531‘ .236 Employment Part-time .517 .583 unemployed. .452 .239 ‘Marital Status .164 .177 Prior Record. .554’" .068 Custodial Status 2.396“* .262 Number of Charges .129 .148 Charge Reduction -.065 .220 Type of Attorney: Self .824’ .398 PUblic Defender .400 .214 Method of Conviction .512 .350 Type of Offense .121 .221 Seriousness .0111" .001 Judge 38 -1.100‘ .526 Judge 75 .205 .492 Judge 112 -.516 .707 Judge 127 .825' .368 Judge 130 -.494 .816 Judge 143 -.471 .660 Constant —2.813"* .489 Log of Likelihood FUnction -407.961 Table 68: Model classification table Predicted 91ft In Out 283 99 Actual In 99 564 Percent Correctly Classified.by Model: 81.05% Percent Correctly Classified.by Chance: 53.68% Percent Reduction in Error Relative to Chance: 59.09% ‘ p < .05 3‘ p < .01 "‘ p < .001 84 effects of other variables included in the model. In particular, the effect of the defendant's race is nearly identical to that reported in Table 4. Therefore, controls for these judges were not included in the remaining analyses of the decision to incarcerate. While the sentencing behavior of individual judges did not add substantially to our understanding of the decision to incarcerate, it is possible that personal characteristics of the sentencing judges might predict this choice. Therefore a model including the race of the judge was estimated. The results reported in Table 7 are similar to the results reported in Table 4. In addition to the variables which were significant in the previous analysis, method of conviction became significant when judicial race was entered into the equation. Those defendants who were convicted after a trial had a higher probability of having a prison term imposed (b=.600, p<.05). Consistent with the bivariate analysis reported in Table 3, the coefficient for the race of the judge was not significant, although it did approach significance (b=-.338, p<.10). The direction of the coefficient suggests that black judges may have tended to be slightly less likely to incarcerate convicted offenders. Possibly the effects of judicial race were masked by the omission of other judicial characteristics. Table 8 reports the results of the estimation of a logit equation 85 Table 7A: Logit estimates of the effects of offender and Juiicial race on the decision to incarcerate, controlling for offense, offender and court case processing characteristics Variable b sge. Offender: Gender .235 .312 Race .163 .232 mmtion - o 586' ‘ o 234 . Enployment mt~time 0494 0364 Unemployed . 429’ . 204 Marital Status .101 .182 Prior Record .544’" .074 Custodial Status 2.429‘" .263 Nunber of Charges .132 .154 Charge Reduction -.122 .204 Type of Attorney: Self .741‘ .371 Public Defender .446‘ .212 Method of Conviction .600‘ .292 Type of Offense .076 .224 Seriousness . 010* ' * . 001 Judicial Race -.338 .186 Constant -2.608"' .490 Log of Likelihood Function -416.663 Table 78: Model classification table Predicted at In Out 283 99 Actual In 99 564 Percent Correctly Classified by Model: 81.05% Percent Correctly Classified by Chance: 53.68% Percent Reduction in Error Relative to Chance: 59.09% * p < .05 " p ( .01 3" p < .001 86 'lable 8A: Logit estimates of the effects of offender and Juiicial race on the decision to incarcerate, controlling for offense, offender, Judicial and court case processing characteristics Vgiable g s.e. Offender: Gender .336 .316 Race . 164 . 234 Education - . 582* . 236 Enployment Part-time .471 .367 Unemployed . 438' . 206 Marital Status .093 .184 Prior Record .540'" .074 Custodial Status 2.409“* .265 Nunber of Charges .144 .157 Charge Reduction —. 155 .206 Type of Attorney: Self .750‘ .374 Public Defender .437‘ .214 Method of Conviction .650‘ .299 Type of Offense .090 .226 Seriousness . 01 1* ‘ * . 001 Judge: Race -.230 .208 Gender -.759‘ .359 Age .001 .011 Tenure .010 .022 commit -2 o 129‘ ‘ ‘ o 719 Log of Likelihood Function -409.270 Table 88: Model classification table meted Quiz 19. Out 280 98 Actual In 98 561 Percent Correctly Classified by Model: 81. 10% Percent Correctly Classified by Chance: 53.71% Percent Reduction in Error Relative to Chance: 59.17% ’ p < .05 n p < .01 3" p < .001 87 including the sex and age of the judge and the number of years on the bench. Again, the estimates produced were similar to those reported in Tables 4, 6 and 7. Race of the judge was not a significant predictor of the likelihood of incarceration, although the direction of the coefficient suggests that black judges may have tended to be less likely to impose a term of imprisonment. Judge's gender, on the other hand, did significantly affect an offender's likelihood of being sentenced to prison. Male judges were less likely to incarcerate defendants than female judges 18 (b=-.759, p<.05) . Another possible explanation for the lack of a racial effect is that race may affect the decision to incarcerate only in cases involving certain racial combinations of judge and offender. For instance, black judges may be less likely 19 to imprison black than white defendants . In order to examine this possibility, an equation including dummy variables representing different racial combinations of judge and offender was estimated. Table 9 presents the descriptive statistics for these variables. As would be expected given the predominance of white judges and black defendants, the majority of the cases were characterized by the white judge-black offender combination (53.3%). Black judges dealt with black offenders in 29.5% of the cases; white judges with white defendants in 11.5%; and the 88 Table 9: Descriptive statistics fer the racial interaction variables included in the logit model (N = 1,045) Variable Min Pg; Mean White judge-black offender 0 1 .538 White judge-white offender 0 1 .115 Black judge—black offender 0 1 .295 Black judge-white offender 0 1 .053 89 remainder were black judges sentencing whites. This last combination served as the category of comparison in the logit analysis. Table 10 reports the results of a logit regression including the interaction variables, offender, offense and case processing characteristics. Again, the estimates produced were similar to those reported in other models. Moreover, none of the racial combinations were significant predictors of case outcome. In order to preclude the possibility that the effects of these variables were suppressed by other judicial characteristics, a model including these factors was also estimated. The results are presented in Table 11. As before, the racial combination of judge and defendant did not significantly affect the likelihood of serving a prison term. It is clear from the analysis presented thus far that initial evidence of racial discrimination against black defendants did not hold up when the effects of other variables were simultaneously controlled. The decision to incarcerate was determined by more legally relevant variables: (1) defendant‘s level of education, (2) employment status, (3) prior record, (4) custodial status, (5) type of defense attorney, (6) the seriousness of the offense, and (7) the method of conviction. Moreover, the race of the judge did not significantly affect this decision. The only judicial characteristic which was 90 Table 10A: Logit estimates of the effects of racial interaction variables on the decision to incarcerate, controlling for offender, offense and court case processing characteristics Variable b s.e. Interactions: White judge-black offender .504 .400 White judge-white offender .344 .458 Black judge-black offender .168 .414 Offender: Gender .235 .312 Education -.586‘ .234 Employment Part-time .494 .364 unemployed .429‘ .204 Marital Status .101 .182 Prior Record .5443” .074 Custodial Status 2.429**‘ .263 Number of Charges .132 .155 Charge Reduction -.122 .204 Type of Attorney: Self .741‘ .371 PUblic Defender .446’ .212 Method of Conviction .600* .292 Type of Offense . 076 . 224 Seriousness .010 .001 Constant —2.950‘** .572 Log of Likelihood Function -416.662 Table 108: ‘Model classification table P l' l l at In Out 283 99 Actual In 99 564 Percent Correctly Classified by Model: 81.05% Percent Correctly Classified by Chance: 53.68% Percent Reduction in Error Relative to Chance: 59.09% * p < .05 ** p < .01 m p < .001 91 Table 11A: Logit estimates of the effects of racial interaction variables on the decision to incarcerate, controlling fer judicial, offender, offense and.court case processing characteristics variable b s.e. Interactions: White judgeéblack offender .395 .410 White judgedwhite offender .232 .468 Black judgeéblack offender .166 .415 Offender: Gender .336 .316 Education .582' .236 Employment unemployed .438' .206 Marital Status .093 .184 Prior Record .540“* .074 Custodial Status .409‘** .265 Number of Charges .144 .157 Charge Reduction .155 .206 Type of Attorney: Self .750‘ .374 PUblic Defender .438‘ .214 Method of Conviction .650‘ .299 Type of Offense .089 .226 Seriousness .011"* .001 Judge: Gender .759‘ .359 Age .001 .011 Tenure .010 .022 Log of Likelihood FUnction -409.270 'lbble 11B: Model classification table Ikedficted Out In Out 280 98 Actual In 98 561 Percent Correctly Classified.by Model: 81.10% Percent Correctly Classified by Chance: 53.71% Percent Reduction in Error Relative to Chance: 59.17% * p < .05 ** p < .01 tn p < .001 92 predictive of sentence outcome was the gender of the judge. Female judges were significantly more likely to sentence a defendant to imprisonment. B. Choices on the Magnitude of the Sentence While racial considerations did not affect the likelihood of imprisonment, they may come into play in deciding the amount of time a defendant will spend behind bars. Descriptive statistics for-the variables included in the analysis of the continuous versions of the dependent variable, length of minumum sentence and proportion of the statutory maximum imposed, are reported in Table 12. This sub-sample of cases was very similar to the larger sample described in Table 1. Approximately 86% of those sentenced to prison were black. This is consistent with the finding that approximately 83% of all defendants convicted were black, indicating that blacks were not over-represented in the group of convicted offenders sent to prison. As in the full sample, nearly a third of these cases were sentenced by black judges. The minimum sentence ranged from one month to 240 months (20 years). The average minimum sentence was 42 months with a standard deviation of 47 months. The proportion of the maximum sentence imposed ranged from a low of .01 to a high of 1.0 which was its limit. In only .5% of the cases did the defendant receive the most severe sentence 93 Table 12: Descriptive statistics for indepeniemt variables incluied in the Ordinary Least Squares analysis (N = 592) Standard variable Min 4gggg ngg; (Deviation Minimun sentence 1 240 42.382 46.584 Proportion of the 0.01 1 .290 .212 legal maximum Offender: Gender 0 1 .944 ---- Race 0 1 .858 -—-- Education 0 1 .137 ---- Employment Part-time 0 1 .073 —-- Unemployed 0 1 .672 --- Marital Status 0 1 .486 ~e-- Prior Record. 0 5 1.443 1.597 Custodial Status 0 1 .529 —-- Number of Charges 0 2 .823 .676 Charge Reduction 0 1 .422 ---- Type of Attorney: Self 0 1 .093 ---- Public Defender 0 1 .704 --—- Method of Conviction 0 1 .182 —--- Type of Offense 0 1 .593 ---- Seriousness 3 300 154.895 123.744 Judge: Gender 0 1 .885 --—- Race 0 1 .324 ---- Age 35 75 50.286 10.394 Tenure 1 42 7.963 5.882 94 defined by the statute. The average defendant received 29% of the legal maximum (s.d.=.212). There were a few differences from the larger sample worth noting. Defendants sentenced to prison were more likely to have been in custody at the time of sentencing (Mean=.529 as compared to .365 in the full sample). In addition, incarcerated offenders were less likely to have had the original charges reduced prior to conviction (.422 vs. .500). Moreover, the crimes committed in the sub—sample were more serious. Nearly 60% of the offenses committed by the incarcerated group were violent crimes, an increase of 10% over the percentage of violent offenses in the full sample. The average statutory maximum sentence was also higher in the smaller sample (154.895 vs. 131.283). It appears, then, that the defendants sentenced to prison had committed more serious violations of the law than those who were placed in the community, a finding consistent with the analysis presented in the previous section. These findings are not surprising, however,since these factors were related to the decision to incarcerate. Table 13 reports zero-order correlations between the race variables and the two continuous versions of the dependent variable. The table demonstrates that neither the race of the offender or the race of the judge was strongly correlated with the minimum sentence imposed on the defendant. The race of the offender was significantly 95 Table 13: Zero-order correlations between race variables and the continuous versions of the dependent variable ARace of the'Offender Race of the Jygge Minimum sentence -.004 -.017 Proportion of the —.070* -.009 legal maximum * p < .05 96 related to the proportion of the maximum sentence imposed, but the association was weak (r=-.07, p<.05). The correlation suggests that black offenders received a slightly lower proportion of the maximum sentence than whites. 1. Analysis of the Length of Minimum Sentence Ordinary Least Squares regression techniques were employed to estimate the effects of offender and judicial race on the minimum sentence in order to simultaneously control for other potentially important variables. Table 14 reports the estimates produced when offense, offender and court case processing characteristics were included in the equation. The results demonstrate that a somewhat different set of considerations came into play in determining the amount of time the defendant was to be incarcerated. The only offender characteristic which significantly affected the length of sentence was prior record. For each past adult criminal conviction, the offender's sentence was lengthened by about 4 months (b=3.829, p<.001). Those‘ defendants who were in custody received sentences which were nearly 10 months longer than those who had been released on bail (b=9.560, p<.001). If the original charge was downgraded, defendants were sentenced to prison terms which were approximately 7 months longer than defendants convicted on the original charge (b=7.104, p<.05). Defendants who 97 Table 14: OLS estimates of the effects of offender race on.the minimum sentence in months, controlling for offender, offense and court case processing characteristics variable b B s.e. Offender: Gender 11.855 .059 6.515 Race -2.026 —.015 4.205 Education -6.019 -.O45 4.379 Employment Part-time -3.071 -.017 6.273 Unemployed 1.281 .013 3.530 Marital Status —1.663 -.018 2.977 Prior Record 3.820“* .133 .963 Custodial Status 9.560" .104 3.121 Number of Charges 3 .640 .054 2 . 357 Charge Reduction 7.104* .076 3.440 Type of Attorney: Self 9.017 .057 5.879 Pablic Defender 2.273 .023 3.852 Method of Conviction 18.798"* .157 3.970 Type of Offense -1.420 -.015 4.328 Seriousness .265*“ .597 .021 Constant -29.062**‘ 8.897 Multiple R = .656 R2 : .431 F = 28.840 ‘ p < 005 83* p < .001 98 were convicted by trial received sentences that were nearly nineteen months longer than those who chose to enter a guilty plea (b=18.798, p<.001). Finally, the seriousness of the offense significantly predicts the length of minimum sentence (b=.265, p<.001). In fact, this variable was the most important predictor of the minimum sentence 20 (B=.597) . This finding should not be surprising and is more of a control than an explanatory factor. The overall 2 fit of the model was a moderately good one (R =.431). Consistent with the results of the zero-order correlation between offender race and minimum sentence reported in Table 13, the effect of offender race was not significant in this model (b=-2.026, p>.10). Although the coefficient suggests that blacks may have received slightly shorter sentences than whites, this difference was in all likelihood due to chance. There is, then, no evidence of racial discrimination. A model examining the effects of individual judges on the length of minimum sentence was also estimated. Descriptive statistics for the eight judges included in the equation are presented in Table 15. As before, only those judges sentencing at least 5% of the cases in the sample were included in the equation as dummy variables; the remaining judges fell in the category of comparison. Similar to the larger sample, no one judge was responsible 99 Table 15: Descriptive statistics for the juiges incluied in the model Percentage Jgge of (h;s___es Race; Gender 15 5.2 W F 17 5.1 W F 42 5.7 W M 75 5.1 W M 83 5.4 W M 127 7.3 W M 130 5.6 W M 143 5.6 B M 100 for imposing sentence in a large proportion of the cases. Two judges, Judge 12 and Judge 17, were female. Only one judge, Judge 143, was black. Table 16 reports the estimates produced as a result of the OLS regression which included the individual judges. The estimates produced were similar to those reported in Table 14. None of the judges significantly affected the length of minimum sentence, although the variable representing Judge 42 did approach significance. One interesting result of this equation concerned the race of the defendant. As in Table 14, the coefficient for this variable was not significant, however, when the judges were added to the model, the coefficient produced was considerably smaller than in the original model. It appears that a portion of the explained variance attributed to race was in fact due to the effects of individual judges, which suggests that the coefficient for defendant race had absorbed variance actually explained by variables previously omitted from the model. Table 17 presents the estimates produced when race of the judge was added to the model. These results were essentially the same as those reported in Table 14. Once again, offender race was not an important predictor of the dependent variable(b=-1.943, p>.10). In addition, the race of the judge did not make a significant impact on the length of sentence(b=-1.166, p>.10), although the direction of the 101 Table 16: OLS estimates of the effects of individual juiges on nurdmmmnsentence in.months, controlling fer offense, offender, and court case processing characteristics variable b _§fi .e. Offender: Gender 12.703 .063 6.566 Race -1.360 -.010 4.247 Education -6.944 -.052 4.442 Employment Part—time -3.577 —.020 6.355 Unemployed 1.107 .011 3.546 Marital Status -2.025 —.022 3.008 Prior Record 3.920*“ .136 .973 Custodial Status 9.0311" .098 3.170 Number of Charges 3.679 .054 2.374 Charge Reduction 7.1461. .077 3.454 Type of Attorney: Self 8.580 .054 5.949 PUblic Defender 2.310 .023 3.895 Method of Conviction 18.958*‘* .159 3.990 Type of Offense -1.702 -.018 4.356 Seriousness .268“‘ .604 .021 Judge 12 1.712 .008 6.743 Judge 17 3.437 .016 6.831 Judge 42 11.237 .057 6.459 Judge 75 10.590 .051 6.848 Judge 127 4.476 .025 5.750 Judge 130 -1.254 -.006 6.538 Judge 143 -1.661 -.008 6.491 Constant -32.024"‘ 9.081 Multiple R = .661 R3 = .437 F = 18.996“‘ * p < .05 3* p < .01 *It p < .001 102 Table 17: OLS estimates of the effects of offender and.judicial race on.minrmum sentence in months, controlling for offense, offender and court case processing characteristics variable jg B :g,e. Offender: Gender 11.736 .058 6.528 Race -1.943 -.015 4.214 Employment Part-time -3.221 -.018 6.291 unemployed 1.236 .013 3.535 Marital Status -1.688 -.018 2.980 Prior Record 3.827"‘ .133 .964 Custodial Status 9.516" .103 3.125 Number of Charges 3.668 .054 2.360 Charge Reduction 7.025‘ .076 3.449 Type of Attorney: Self 8.903 .056 5.892 PUblic‘Defender 2.256 .022 3.855 Method of Conviction 18.830*"“t .157 3.974 Type of Offense —1.421 -.015 4.331 Seriousness .265‘*‘ .596 .021 Judical Race —1.166 -.012 3.128 Constant -28.524**‘ 9.100 Multiple R = .656 R3 = .431 F = 27.006 * p < .05 3* p < .01 tit p < .001 103 coefficient suggests that black judges may have tended to give slightly shorter sentences. In Table 18 the equation including judicial characteristics other than race is presented. Again, the estimates were similar to those found in Tables 14, 16 and 17. None of the characteristics of the judge included in the model appear to have had a significant impact on the length of the minimum sentence imposed on the defendant. This finding is noteworthy since the analysis of the effects of judicial characteristics on the decision to incarcerate (Table 6) indicated that female judges were more likely to sentence defendants to terms of imprisonment. It appears, then, that while offenders sentenced by female judges were slightly more likely to be incarcerated, the length of their sentences was not significantly different from those imposed on defendants sentenced by males. Table 19 presents the descriptive statistics for variables representing the different racial combinations of judge and defendant. The data in this sub-sample were similar to those described in Table 9. The majority of the cases involved white judges sentencing black defendants (57.1%). The least likely combination was a black judge sentencing a white offender. Only 3.7% of the cases fell into this category, therefore, these cases were excluded form this portion of the analysis, leaving a sample of 570 cases. The racial combination of black judges dealing with 104 Table 18: OLS estimates of the effects of offender and judicial race on the minimun sentence, controlling for offender, offense, judicial and court case processing characteristics variable .9 jg gg§.e. Offender: Gender 11.158 .055 6.525 Race -1.340 -.010 4.216 Education -5.969 -.045 4.388 Employment Part—time -3.467 -.020 6.291 unemployed 1.251 .013 3.531 Marital Status -1.961 —.021 2.979 Prior Record. 3.792‘*‘ .132 .965 Custodial Status 8.965“I .097 3.138 Number of Charges 4.006 .059 2.363 Charge Reduction 7.005‘ .075 3.449 Type of Attorney: Self 9.722 .062 5.911 PUblic Defender 3.061 .030 3.873 Method of Conviction 18.719*" .157 3.971 Type of Offense —1.520 -.016 4.330 Seriousness .268*‘* .603 .021 Judge: Race —2.040 -.021 3.374 Gender —4.270 —.030 5.188 Age .031 .007 .188 Tenure -.483 —.062 .320 Constant -22.971 11.987 Multiple R.= .660 R? = .436 F = 23.064 * p < .05 *x p < .01 xxx p < .001 105 Table 19: Descriptive statistics for the racial interaction variables included.in the OLS models (N = 592) variable Mfin. ngx Mean White judge-black offender 0 1 .571 White judge-white offender 0 1 .105 Black judge-black offender 0 1 .287 Black judge-white offender 0 1 .037 106 black offenders served as the category of comparison in the regression equation. The results of the estimation of this equation are reported in Table 20. As in the case of the decision to incarcerate, specific racial combinations did not significantly affect the length of minimum sentence. Inclusion of other judicial characteristics in the equation also failed to reveal a racial effect. (See Table 21 for the results of this analysis.) The analysis presented in this section indicated that racial considerations did not affect the decision to impose a specific minimum sentence on defendants. Moreover, the characteristics of judges did not affect the dependent variable. The choice of minimum sentence appears to have been shaped primarily by the custodial status and prior record of the defendant, the method of conviction and the seriousness of the offense. 2. Analysis of the Proportion of the Statutory Maximum Imposed A final set of regression equations was run in order to determine if racial considerations affected the proportion of the legal maximum sentence imposed on the defendant. Equations estimating these effects contained the same variables related to offender, offense and court case 107 Table 20: OLS estimates of the effects of racial interactions on minimun sentence, controlling for offender, offense and court case processing characteristics variable b B s.e. Interactions: White judge-black offender 2.636 .028 3.363 White judge-white offender 1.330 .009 5.318 Offender: Gender 9.889 .048 6.864 Education -6.516 -.049 4.490 Employment Part-time -2.286 -.013 6.457 Unemployed .941 .010 3.605 Marital Status -1.150 -.013 3.048 Prior Record 3.871‘33 .135 .995 Custodial Status 9.781" .106 3.218 Number of Charges 3.927 .058 2.421 Charge Reduction 6.659 .072 3.548 Type of Attorney: Self 8.857 .056 6.069 Pablic Defender 1.562 .016 3.950 Method of Conviction 19.190*“ .159 4.108 Type of Offense -1.703 -.018 4.411 Seriousness .263*‘* .593 .022 Constant o30.221*‘* 8.802 Multiple R = .653 R2 = .426 F = 25.450*‘* 3 p < .05 it p ( .01 xx: p < .001 'lable 21: 013 estimates of the effects of racial interaction variables 108 onlmdnimum.sentence, controlling for judicial, offense, offender and.court case processing characteristics variable b B s.e. Interactions: White judge-black offender 3.835 .041 3.612 White judge-white offender 1.673 .011 5.446 Offender: Gender 9.187 .044 6.859 Education -6.450 -.048 4.482 Employment Part-time 2.421 .013 6.455 Unanployed . 957 . 010 3 . 599 Marital Status -1.390 —.015 3.045 Prior Record. 3.820" .133 .996 Custodial Status 9.185** .100 3.231 Number of Charges 4.298 .063 2.424 Charge Reduction 6.633 .071 3.547 Type of Attorney: Self 9.737 .061 6.082 Public Defender 2.516 .025 3.967 Method of Conviction 19.100*** .158 4.102 Type of Offense —1.969 —.021 4.415 Seriousness .266*“ .600 .022 Judge: Gerlder -3 0694 “0026 5028]. Age .051 .112 .191 Tenure -.556 -.071 .325 Constant -26.049* 11.781 Multiple R.= .657 R? = .431 F = 21.7863*' 3 ‘p < .05 it p ( .01 #13 p < .001 109 processing characteristics as previously discussed analyses, with one exception. The variable indicating the seriousness of the offense was not included as an independent variable since it was used to calculate the dependent variable. If the variable were to be included in the model, the estimated relationship between it and the dependent variable would be so overwhelming as to mask the true effects of the other variables in the model. Table 22 reports the results of an OLS regression in which the effects of the race of the offender, as well as other offender, offense and court case processing characteristics, were estimated. In this model, defendants with longer criminal histories received more of the statutory maximum (b=.029, p<.001). In contrast to analysis of the decision to incarcerate and the length of minimum sentence, prior record was the most influential factor in determining the proportion of the maximum imposed (B=.219). Defendants who were in custody received a larger proportion of the legal maximum than their counterparts who made bail (b=.056, p<.001). The number of charges brought against the defendant also significantly increased the proportion of the maximum imposed on the defendant (b=.035, p<.01). In the models estimating the effects on the other versions of the dependent variable the effect of the number of charges never attained significance, although in the models with minimum sentence as the dependent variable this variable did Table 22: OLS estimates of the effect of offender race on the proportion of the legal.maximum imposed, controlling fer offense, offender and court case processing llO characteristics variable :9 B g§.e. Offender: Gender .018 .019 .037 Race -.025 —.041 .024 Education -.027 -.043 .025 Employment Part-time -.010 -.012 .035 unemployed .017 .038 .020 Marital Status .018 .043 .017 Prior Record .029"' .219 .005 Custodial Status .056**’ .132 .017 Nunber of Charges .035" .112 .013 Charge Reduction .037‘ .087 .018 Type of Attorney: Self .074’ .103 .033 Pablic Defender .013 .029 .022 Method of Conviction .097'** .176 .022 Type of Offense -.050**‘ -.117 .018 Constant .158*'* .049 Multiple R : .379 R2 = .144 F - 6.860 ‘ p < .05 *t p ( .01 lll approach significance (p<.10; See Tables 14-21). As before, cases in which the original charge was reduced received a larger proportion of the maximum sentence (b=.037, p<.05). Defendants who attempted to represent themselves in court also received a larger percentage of the maximum (b=.074, p<.05) than those offenders represented by private attorneys. Also, those defendants who were tried had a larger proportion of the maximum imposed (b=.097, p<.001). Interestingly, violent offenders received a lower proportion of the maximum sentence (b=-.050, p<.001). This may be due to the fact that many violent offenses are crimes of passion, not involving premeditation. Such offenders may not receive sentences that are close to the legal maximum if they are not expected to offend again. In addition, there is a greater range of possible sentences available for violent offenses, thus allowing judges more leeway in determining the proportion of the maximum to impose on the offender. The effect of this variable may have been overwhelmed by the variable measuring offense seriousness in other analyses. The race of the offender did not significantly affect the proportion of the maximum sentence imposed on the offender (b=-.025, p>.10). The coefficient indicated that black offenders received a smaller proportion of the legal maximum, but this was probably due to chance. Thus, the weak association between race and this dependent variable 112 noted in Table 13 disappeared when the analysis controlled for the effects of other variables. The goodness-of-fit for this model was not nearly as good as that found in models 2 discussed previously (R =.144). In the model including the eight individual judges (see Table 23), Judge 75 was found to contribute significantly to the proportion of the maximum imposed (b=.099, p<.01). Defendants who were sentenced by this judge recieved 10% more of the statutory maximum than those whose cases were handled by other judges. The remainder of the coefficients were consistent with those reported in Table 22. In the model which included the race of the judge (see Table 24), the estimates produced by regression techniques were essentially the same as those reported in Tables 22 and 23. As expected based on the zero-order correlation of judge's race with this version of the dependent variable, the race of the judge did not significantly affect the proportion of the legal maximum imposed on the defendant (b=.009, p>.10). In addition, the coefficient for offender race remained non-significant (b=-.025, p<.10). Table 25 reports the estimates of the model which added other judicial characteristics to the equation. These results were also very similar to those reported in Tables 22, 23 and 24. The variables measuring the race of the judge and offender did not significantly affect the dependent 113 Table 23: OLS estimates of the effects of individual judges on the proportion of the legal maximum imposed, controlling for offense, offender and court case processing characteristics variable b g; s.e. Offender: Gender .028 .030 .037 me ‘0017 -0028 0024 Education -.029 -.047 .025 Employment Part-time -.007 -.008 .035 Unemployed .018 .040 .020 Marital Status .013 .032 .017 Prior Record .031"‘ .231 .005 Custodial Status .055" .131 .017 Number of Charges .033“ .105 .013 Charge Reduction .038‘ .088 .018 Type of Attorney: Self .066‘ .091 .033 Pablic Defender .011 .024 .022 Method of Conviction .095‘*‘ .181 .022 Type of Offense -.049** -.114 .018 Judge 17 .032 .033 .038 Judge 42 .041 .045 .036 Judge 75 .099" .103 .038 Judge 127 .012 .015 .032 Judge 130 -.041 -.045 .037 Judge 143 —.041 -.045 .036 Constant --.141" .050 Multiple R = .405 R2 = .164 F = 5.041*" ‘ p < .05 xx p < .01 it! p < .001 ll4 'lable 24: OLS estimates of the effects of offender and juiicial raceontheproportionofthelegalmxinnimposed, controlling for offense, offender and court case processing characteristics Variable b B s.e. Offender: Gender . 019 .020 . 037 Race . 025 - . 042 . 024 mtim o 026 - o 042 o 025 Enployment Part-time . 009 - . 011 . 035 Uneuployed . 017 . 039 . 020 Marital Status . 018 . 043 . 017 Prior Record . 029‘ * * . 219 . 005 Custodial Status .056‘" . 133 .017 Nunber of Charges .035" .111 .013 Charge Reduction . 038* . 088 . 018 Type of Attorney: Self . 075‘ . 104 . 033 Public Defender . 013 . 029 .022 Method of Conviction . 096’ 1' ’ . 17 5 . 022 Type of Offense .050‘" -.117 .018 Judicial Race . 009 . 019 .018 Constant . 154'I ‘ . 050 Multiple R = .379 R2 = .144 F = 6.411 * p < .05 3* p < .01 it: p < .001 115 Table 25: OLS estimates of the effects of offender and.judicial race on the proportion of the legal maximum imposed, controlling for offense, offender, judicial and.court case processing characteristics m p < .001 Variable Q E s.e. Offender: Gender .015 .016 .037 Race -.022 —.036 .024 Education -.025 -.040 .025 Employment Part-time -.008 -.010 .035 Unemployed .017 .038 .020 Marital Status .016 .039 .017 Prior Record .028333 .215 .005 Custodial Status .05533 .129 .017 Number of Charges .0373 .117 .013 Charge Reduction .03733 .086 .018 Type of Attorney: Self .0773 .107 .033 Public Defender .017 .037 .022 Method of Conviction .096333 .175 .022 Type of Offense -.04933 -.115 .018 Judge: Race -.001 —.003 .019 Gender -.002 -.003 .029 Age .0003 .015 .001 Tenure -.0043 -.103 .002 Constant .17233 .067 Multiple R.= .390 R3 = .152 F = 5.679 3 p < .05 it p < .01 ll6 variable. The number of years served on the bench by the judge did predict the outcome, however. The percentage of the maximum imposed by a judge was decreased by .4% for every year he/she had served on the bench (b=-.004, p<.05). It appears that the judges became more lenient the longer their tenure or that the more recently elected judges were more conservative. The fit for this model was somewhat better than that found for the first two equations examining 2 this dependent variable (R =.152). Tables 26 and 27 report the estimates produced by the equations including the racial combinations of judge and defendant. These results were consistent with other analyses of this dependent variable. In addition, the racial combination variables did not contribute significantly to the variance explained by the model. In sum, the proportion of the maximum imposed on the defendant was affected by the same kinds of considerations as those that contributed to the minimum sentence, i.e., custodial status, prior record, the number of charges, the reduction of charges by the prosecutor, type of offense, and method of conviction. However, these factors explain only a small portion of the variance in the dependent variable. This may have occurred for three reasons. First, it is possible that relevant factors were ommitted from the model. The proportion of the maximum sentence imposed on 117 Table 26: OLS estimates of the effects of racial interaction variables on.the proportion of the legal maximum imposed, controlling for offense, offender and.court case processing characteristics variable 'b B s.e. Interactions: White judgeeblack offender -.008 -.018 .019 White judge-white offender .014 .020 .030 Offender: Gender .002 .002 .039 Education —.025 -.042 .025 Employment Unemployed .014 .030 .020 Marital Status .019 .045 .017 Prior Record .028333 .211 .006 Custodial Status .062333 .145 .018 Number of Charges .0343 .110 .014 Charge Reduction .0383 .090 .019 Type of Attorney: Self .0803 .109 .034 Public Defender .011 .023 .022 Method of Conviction .095333 .171 .023 Type of Offense —.05233 -.122 .019 Constant .15633 .049 Multiple R = .372 R2 = .138 F = 5.881333 3 p < .05 *t p < .01 xxx p < .001 118 Table 27: OLS estimates of the effect of the racial interaction variables on the proportion of the legal maximum imposed, controlling fer judicial, offense, offender and court case processing characteristics variable 'b B 44§.e. Interactions: White judge4black offender .004 .008 .020 White judge—white offender .020 .030 .031 Offender: Gerder " o 002 "' o 002 o 039 Education -.025 -.041 .025 Employment Unemployed .014 .030 .020 Marital Status .018 .042 .017 Prior Record .027333 .206 .006 Custodial Status .0603"3 .141 .018 Number of Charges .03633 .116 . .014 Charge Reduction .0373 .087 .019 Type of Attorney: Self .0823 .112 .034 Public Defender .015 .032 .022 Method of Conviction .095 .171 .023 Type of Offense -.052 -.121 .019 Judge: Gender .002 .003 .030 Age .0004 .018 .001 Tenure —.0043 -.105 .002 Constant .16033 .066 Multiple R = .383 R2 = .146 F = 5.213333 3 p < .05 it p < .01 Mt p < .001 119 the defendant may depend on factors which have not been considered in the present analysis, e.g., the judge's beliefs about punishment, retribution, or deterrence. Thus, the variable may be better understood in terms of factors that are as yet unknown. Second, it is possible that this aspect of the sentencing decision does not follow a regular pattern. In this instance, the model might be correctly specified and still explain only a small portion of the variation in the dependent variable. Third, analyses of other versions of the sentencing decision may have inflated estimates of explained variance because offense seriousness is used as a predictor variable. If sentences are designed by the legislature on the basis of offense seriousness, statistical models utilizing this factor as a predictor may overestimate the variance explained by the independent variables in the model. By placing offense seriousness in 2 the dependent variable, it is possible that the low R obtained is a more accurate estimate than those obtained in previous models. Further research is necessary to clarify these issues. C. The Process of Imposing Sentence Gibson (1983) noted that the outcome of the decision making process may be similar, while the process itself 120 varies across groups. That is, even though judicial race did not significantly affect the outcomes of sentencing examined in the present study, it is possible that black and white judges consider different factors when reaching their decisions. In order to investigate this possibility, a series of regression equations was estimated separately for black and white judges. Table 28 presents the coefficients produced when the model including offender, offense, and case processing characteristics was estimated for the decision to incarcerate. For black judges, the data indicated that defendants with longer criminal records were more likely to be sent to prison (b=.438, p<.001). Those who were unable to make bail were also more likely to be incarcerated (b=1.807, p<.001). In addition, the seriousness of the offense contributed to the likelihood of incarceration (b=.010, p<.001). As expected in light of the previously discussed analyses, the race of the defendant did not significantly affect this dependent variable. The model fit the data well; the percent improvement relative to chance was 53.98%. The decisions of white judges, on the other hand, were affected by more factors. Defendants with at least a high school education were less likely to receive a prison sentence than those who did not complete high school (b=-.816, p<.001). Unemployed offenders were significantly lj21. Table 28: Comparisons of the factors affecting the decision to :irrcasrraezr1ntue» £231. tilunrik: suns! Iairistse ;iuriirens (Iroirijb) Cluck lye: git: 1m Viriuhle h II b In Offender: Sex .068 .465 .415 .427 Race .210 .491 .187 .295 Education -.395 .391 -.816"‘ .301 Rlploylent Part-tile .774 .750 .412 .465 Unelployed .356 .314 .5174 .262 Marital Status -.014 .279 .136 .224 Prior Record .438"‘ .099 .665!!! .109 Custodial Status 1.807m .371 3.036" .413 Nulber of Charges -.031 .261 .211 .200 Charge Reduction -.057 .328 -.177 .261 Defense Attorney: Self .381 .837 .992‘ .461 Public Defender .132 .353 .630‘ .264 Method of Conviction -.126 .404 1.419“‘ .429 Type of Offense .148 .395 -.008 .280 Seriousness .010m .002 .012m .002 Constant -2.087m .785 ~3.296m .645 Log of the Likelihood Function: -165.239 Log of the Likelihood Function: -241.267 Percent Correctly Classified by Model: 77.96% hmwthnmfiyfluflfidbywmmz52fl3 hmmthmwnmtkhfintowmm:5i%1 Percent Correctly Classified by Model: 82.843 hmwtknwdyfluflfifibyflmmz5LWX Percent Ilprovelent Relative to Chance: 62.983 : 32 p ( .05 p ( .01 xx: p ( .001 122 more likely to be imprisoned than defendants who were working full time (b=.517, p<.05). Prior record contributed to the likelihood of incarceration (b=.665, p<.001). Defendants in custody at the time of sentencing were also more likely to be sentenced to a period of imprisonment (b=3.036, p<.001). The type of defense attorney was also an important predictor of case outcome. Defendants represented by public defenders and those attempting to represent themselves were more likely to be imprisoned (b=.630 and b=.992, respectively, p<.05). The seriousness of the offense also contributed to case disposition (b=.012, p<.001). This model also fit the data well; the percent improvement relative to chance was 62.98%. Clearly, different factors were important in shaping the decisions of black and white judges. White judges appear to have been influenced by more extra-legal factors than black judges. They considered the education and employment status of the defendant in deciding whether incarceration would be appropriate. They also appear to be more susCeptible to differences in the type of attorney presenting the defense. It is possible that white judges are more likely to discriminate against offenders on the basis of economics than black judges. Black judges appear to rely almost solely on legally relevant factors for this decision. Table 29 reports the OLS equations analyzing the flbilnlsee 7254: 14213 usenitxanlae: thor' lilrncdr tlrli.‘wdiifitse ,jtsdhgens ((1143) ‘(lfllllplllfiilitlfllB (>1? itines lisusdhrrrus. sniffisxz1risnlz itine: lluarnirtflhl (>1? llElIliJlIllI Ilse“ es lite rm Vuriuhle h 8 a b I a Offender: Sex 13.658 .081 10.225 13.119 .059 8.655 Pace -9.530 -.072 7.896 1.462 .011 5.103 Sducstion -5.631 -.042 8.091 -7.652 -.058 5.319 Siploylent Part-tile 11.797 .065 11.549 -9.862 -.056 7.665 Unelployed 9.115 .102 5.787 -1.931 -.019 4.511 Marital Status 1.025 .012 5.053 —2.342 -.025 3.745 Prior Record 1.639 .064 1.555 5.148111 .169 1.267 Custodial Status 6.897 .082 5.277 9.554“ .100 3.945 Mulber of Charges 1.728 .029 3.893 4.717 .066 3.028 Charge Reduction 7.610 .088 6.012 6.669 .070 4.319 Defense Attorney Self 17.814 .110 10.794 4.343 .028 7.229 Public Defender 2.996 .032 6.406 1.088 .010 4.955 Method of Conviction 18.019" .168 6.558 18.510“‘ .147 5.103 Type of Offense 3.830 .045 7.241 -2.813 -.029 5.471 Seriousness .223"' .546 .037 .284m .616 .026 Constant -23.902 14.636 -32.861" 11.524 Multiple R : .650 Multiple R : .672 111 = .422 p: = .451 F = 8.575“‘ F = 20.812331 1 p ( .05 1: p < .01 xx: p < .001 124 effects of variables in the model on the length of minimum sentence for black and white judges. Very few factors influenced this decision for black judges. Defendants who were convicted after a trial received sentences that were slightly more than 18 months longer than those who entered a plea of guilty (b=18.019, p<.01). In addition, the seriousness of the offense added to the offender's sentence (b=.223, p<.001). This model fit the data well, explaining slightly more 42% of the variance in the dependent variable. White judges were again influenced by more factors in this decision. The length of the minimum sentence was increased by 5 months for every adult felony conviction in the defendant's criminal history. Being in custody at the time of sentencing increased the offenders sentence by nearly 10 months. Conviction at trial resulted in an increase of 18.5 months. The seriousness of the offense also increased the length of minimum sentence. This model also fit the data well, explaining 45% of the variance in the dependent variable. Table 30 presents the results of the regression analysis of the proportion of the statutory maximum received by the defendant. As in the case of minimum sentence, black judges are influenced by the method of conviction. Offenders who go to trial received nearly 15% more of the maximum sentence possible than those who entered a plea of 15255 'lable 30: Camarisons of the factors affecting the proportion of the estusibirtznuryr Illusrinuusnl :innpacuauasi 'fiaur ‘tilnarzlr terns! latiistaer ;)\ldhlless ((thS!) Block 1 s Ihite Igglgs Vuriuhle b 6 n h I h Offender: sex ‘0047 -0057 0080 0049 0049 0041 Race -.028 -.043 .046 -.021 -.036 .028 Education -.034 -.051 .048 -.027 -.046 .029 Riploylent Part'ti.” 0019 0089 6068 -0041 '0061 0041 One-played .059 .134 .033 -.008 -.019 .024 Marital Status .030 .073 .030 .017 .039 .020 Prior Record .016 .129 .009 .032133 .239 .007 Custodial Status -.021 -.049 .031 .089"' .208 .021 Mulber of Charges .035 .118 .023 .0391 .121 .016 Charge Reduction .012 .028 .032 .038 .090 .022 Defense Attorney Self .070 .087 .063 .076 .110 .039 Public Defender .063 .137 .038 -.003 -.007 .027 Method of Conviction .149“ .282 .039 .077"' .137 .028 Type of Offense -.D45 -.105 .031 -.051‘ -.118 .023 Constant .208" .084 .132' .061 Multiple R = .410 Multiple P = .428 9' = .166 8' = .183 F = 2.559333 P = 6.102"' 1 p ( .05 x: p < .01 m p < .001 126 guilty. White judges, on the other hand, handed down more of the maximum sentence when the defendant had been convicted of other felonies in the past (b=.039, p<.001). Defendants received a greater proportion of the maximum sentence if they were in custody (b=.089, p<.001) and if they were facing several charges (b=.039, p<.05). Conviction at trial also increased the proportion of the maximum imposed on the defendant (b=.077, p<.001). Interestingly, offenders convicted of violent offenses received a smaller proportion of the statutory maximum. As in the case of the model estimated on the full sample of incarcerated offenders, neither model explains much variance in the dependent variable. Overall, it appears that while sentencing patterns were not influenced by the race of the judge, different factors did come into play in the process of making the decision. Black judges appear to have relied on a few legally relevant factors. In comparison, white judges allowed more kinds of variables to influence their decisions. The models estimated appeared to fit the data for white judges better than for black judges, but the difference is small. It is possible that factors ommitted from the model affected the decisions of black judges. However, since the-amount of variance explained by the models was not substantially different for black and white judges, it is reasonable to conclude that black judges simply rely more heavily on a few 127 variables. It appears, then, that the process of decision making for black and white judges differs even though the average outcome is the same. CHAPTER V DISCUSSION AND CONCLUSIONS The analysis presented here failed to find any evidence of racial discrimination in the sentencing of felony offenders. Initial indications of disparity in the decision to incarcerate and the proportion of the statutory maximum sentence imposed disappeared in equations which simultaneously controlled for the effects of other, more legally relevant variables. Based on these findings, then, one can tentatively conclude that sentencing decisions were made independently of the race of the offender. Further, consistent with Uhlman's (1977) work, black judges did not impose sentences which were significantly different from those handed down by their white counterparts, although the direction of the coefficient did suggest that they may have tended to be more lenient. This was true even when offender, offense, court case processing and other judicial characteristics were simultaneously controlled. These results are consistent with the expectations generated by the consensus approach, i.e., that black defendants would not receive more severe sentences and black judges would impose sentences similar to those imposed by their white counterparts. The finding of no racial effect appears to confirm the assumption of beliefs that are 128 129 universally shared by the members of society. Not only do our laws forbid felonious behavior, but our legal officials find such acts equally repugnant regardless of their own or the defendant's minority status. Before we accept the consensus perspective, however, there are a few caveats that must be noted. First, due to limitations in the data, the present study only examined the issue of direct racial discrimination, i.e., unequal treatment directly attributable to the race of the defendant. It is possible that discrimination in sentencing is of a less noticeable form. For instance, there is evidence based on research examining the imposition of the death penalty that the victim's race may play a significant role in the sentencing decision (Radelet & Pierce, 1985; Bowers, 1983; Gross & Mauro, 1984; Baldus et al., 1983a, b, 1985). These studies have consistently demonstrated that killers of white victims are more likely to be charged with a capital offense and to have the death penalty imposed than killers of blacks. Moreover, black killers of whites are more likely than any other defendant-victim racial combination to receive a death sentence, while black killers of blacks are the least likely group to face death row. These studies have indicated that the life of a white victim is more valuable than the life of a black in our legal system, a conclusion which is consistent with the conflict View. While decisions to impose capital punishment are 130 generally made by juries, it is possible that the same type of discrimination occurs in the sentencing decisions made by judges. The Subtlety of discrimination by the race of the victim is also compatible with this approach. Conflict theorists have often observed that the state must disguise the true nature of the law. For instance, Quinney suggested that the groups in power utilize the media to construct a social reality that allows members of the oppressed group to define behavior which is in opposition to the interests of more powerful groups as criminal (1970). In this way the laws are accepted as legitimate by groups whose interests are being thwarted by those same laws, thereby perpetuating unequal treatment in society. Discrimination on the basis of the victim's race would be hidden from the View of the layman. Thus, it would serve the purpose of protecting dominant group interests while simultaneously leading the less powerful to believe, albeit falsely, in the equity of American law. It is very possible that this more subtle form of discrimination would be found in the present study, if adequate information was available. Unfortunately, the data did not allow an analysis of this question. Therefore, support for the consensus approach on the basis of a finding of no direct discrimination can only be tentative. 131 A second caveat concerns similarities in the backgrounds of black and white judges. Uhlman (1977) reported that the social backgrounds of black judges are nearly indistinguishable from those of white judges. Chambliss and Seidman (1971) have suggested that judges come from the more privileged segments of society, are trained in law schools by the "casebook" method, and must learn to deal appropriately with the politically powerful. All of these characteristics tend to orient the values of judges toward the protection of the interests of the wealthy. If black and white judges are similar in background and training, black judges may not identify with other blacks as a minority group. Rather, they may be more closely identified with white judges due to the similarities in their backgrounds. In this case, black judges would not deal with blacks or other socially disadvantaged defendants more leniently than their white counterparts. The tendency for discrimination to fall along lines of social class rather than race may have been obscured in the present study for several reasons. First, this study focused on crimes that were primarily of the garden variety, e.g., robbery, homicide, assault and burglary. These offenses are typically committed by members of the lower class, thus limiting the variation in social class in the sample. This problem is compounded by the lack of a solid measure of socioeconomic status in the data. Second, as 132 Hagan and Bumiller (1983) observed, in large bureaucratic 7 courts faced with large caseloads and pressure for efficiency, judges may not be able to individualize the treatment of defendants to the extent that judicial values could significantly influence case disposition. A third caveat cconcerns the analysis of the weights black and white judges placed on factors influencing the sentencing decision. The analysis clearly showed that black judges relied on factors that were more legally relevant, while whites considered a broad range of factors, including those that were related to the defendant's social position. This finding could be interpreted as indicating that the penal philosophy of black judges is oriented towards deterrence and retribution, whereas white judges are more rehabilitative. It could also be taken to suggest that white judges are more susceptible to a form of discrimination which is based on economics. That is, they tend to be more severe with defendants who cannot muster the resources to protect their interests adequately. This would be consistent with the conflict hypothesis that white. judges, as members of the dominant group, would be more severe with the economically disadvantaged because such individuals are perceived as a threat to the wealthy. One of the limitations of conducting a secondary analysis is that the data collected were not intended to measure concepts of interest to the present study. Thus, 133 some concepts must be measured indirectly through indicator variables. This reliance on even more imperfect measures than usual decreases the certainty with which a specific interpretation can be attributed to a particular variable. For instance, the data collected by Zalman et a1. did not contain a measure of socioeconomic status. Therefore employment status, amount of education and marital status were used as indicators of the defendant's SES. Unfortunately, judges who rely on these factors in their sentencing decisions may view them as indicators of the offender's ties to the community and, therefore, his potential for rehabilitation, rather than measures of social class. Without additional information on the penal philosophies of the judges serving the court and the interpretations they give to factors considered in sentencing, it is impossible to determine with any degree of certainty whether judges are discriminating against lower class offenders or merely following a rehabilitative penal philosophy. There is, then, yet another reason to reach somewhat tentative conclusions. The caveats noted here all concern questions that need to be addressed empirically. Analysis that includes consideration of a more extensive range of factors than was possible here (e.g., race of victim, SES, white collar offenses, organizational constraints and the penal philosophies of individual judges) would be required to rule 134 out these possibilities. At this point, however, the data appear to support the consensus view. The analysis indicated that the gender of the judge significantly affected the defendant's chances of being incarcerated. Male judges were found to be less likely to sentence offenders to terms of imprisonment. The results of analyses examining minimum sentence and proportion of the maximum sentence were non-significant for this variable, however, the coefficients suggested that males impose shorter minimum sentences and less of the statutory maximum. The finding of an effect for the decision to incarcerate is consistent with results reported by Gruhl and his colleagues (1981) and, in the case of larceny, Kritzer and Uhlman (1977). These authors found their results to substantively weak, however. In order to examine the substantive strength of this effect, the odds multiplier for the effect was calculated by taking the antilog of the coefficient. Defendants sentenced by male judges were approximately half as likely to receive a prison sentence (.468) as those sentenced by female judges. Thus the effect of judicial gender on the likelihood of incarceration is fairly substantial. The finding that female judges are more likely to impose a term of incarceration is inconsistent with the expectations derived from the conflict perspective. Based on that approach, female judges were expected to be more 135 lenient in their decision than male judges due to their experiences as members of a disadvantaged minority group. This finding is more compatible with the suggestion presented by Gruhl et a1. (1977) that women may be more threatened by challenges to social norms than men due to differences in the socialization process. In particular, obedience, passivity and docility are emphasized for females, while aggressiveness is promoted in males. Thus, women would be more sensitive to violations of the law which, in turn, would lead them to impose more severe sentences than men. One other judicial characteristic was found to have a significant impact on the sentencing decision, the number of years served on the bench. This variable approaches significance in the model predicting minimum sentence (see Table 18) and attains significance when the proportion of the maximum sentence is examined (Table 25). The coefficients suggest that the longer the judge has served on the bench the less he/she is inclined to impose long sentences on offenders. This result may be due to the way in which judges are socialized by their peers when they first begin their careers. That is, judges who have been on the bench for many years may have a different view of the appropriate length of the minimum sentence than those who have more recently entered their circle. This idea is consistent with Gibson's suggestion that variables related 136 to the socialization of judges are the most important in predicting their sentencing behavior (1983). A second possibility is that new types of judges are being elected, e.g., voters may be electing more conservative judges than they had in the past. The variables which most consistently predicted the sentencing outcome are (1) method of conviction, (2) prior record, (3) custodial status, and (4) seriousness of offense. This finding is consistent with the results of other sentencing studies (Blumstein et al., 1983). One interesting result was connected with the charge reduction variable. Contrary to what one might expect, defendants for whom the original charge was reduced received significantly longer sentences and a higher proportion of the legal maximum than those who were convicted on the original charge. It appears, then, that judges are sentencing defendants on the basis of their actual behavior rather than on the conviction charge. This result is consistent with the work of Wilkins et al. (1978) who reported that judges focus on the actual harm caused to the victim in determining a sentence, regardless of the plea entered or the official charge. Thus, judges are considering the harm inflicted by the act, compensating for the effects of the plea bargain. This finding is also similar to results reported by Hagan (1974). His examination of the effects of charge alteration revealed that reducing the charges significantly decreased 137 the probability of conviction, but did not affect the magnitude of the sentences imposed on convicted defendants. Here, charge alteration increased the magnitude of the sentence imposed on the offender relative to individuals convicted of similar offenses with no charge reduction. In addition to empirical support for this finding, there is case law which indicates that judges have the right to consider the actual harm caused by the defendant's actions. In §t§te 3. Henry, the Maryland court of appeals upheld the imposition of the sentence imposed on a defendant for larceny and receipt of stolen goods. The case involved the use of a stolen car in the commission of a robbery and homicide. The petitioner was convicted for the theft of the car and receiving $16 in stolen cash, for which he received the maximum sentence for each charge to run consecutively. Henry appealed on the grounds that the bases used by the trial judge in imposing the sentence violated his due process rights under the 14th Amendment. Specifically, Henry argued that it was improper for the judge to consider his involvement in the robbery and homicide for which he had been acquitted. In upholding the sentence, the appellate court stated: In passing sentence the trial judge was not required to remain oblivious to evidence of Henry's involvement in the homicide and robbery at a level less than would warrant his conviction of those crimes (1974: 305). 138 In a similar case, State v, Sharpe, the appellant entered a plea of guilty on a charge of importation of heroin in exchange for which the prosecutor dropped charges involving the transportation of heroin and two counts of conspiracy to commit murder. Sharpe was sentenced to serve a period in prison which was no less than 45 years and no more than 75 years. He contended in his appeal that the sentence was unduly excessive given the offense for which he was convicted. The Arizona court of appeals held that: The sentencing judge can exercise wide discretion in the sources and types of evidence used to assist him in determining the kind and extent of punishment to be imposed within the limits fixed by law. The court may, before imposing sentence, consider both the nature and circumstances of the crime charged and general background of the defendant (1975: 413). - Therefore, the court ruled that the trial judge properly considered the total circumstances of the offense in determining the sentence to be imposed. This finding is of some import for those who argue that charge bargaining causes the criminal justice system to be more lenient on offenders than they actually deserve. It appears that, in actuality, defendants for whom the official charge is reduced received sentences that were longer than others who were convicted without charge alteration. Moreover, this reaction on the part of sentencing judges has been upheld in case law. Thus, the benefits of charge negotiations accrued by defendants may be more illusory than 139 real. Racial considerations, then, do not appear to play an important role in the sentencing decisions of either white or black judges. This is a significant finding, especially for those who believe that the inclusion of more black judges on the bench will serve to eliminate--or at least alleviate--the problem of racial discrimination. Based on these data, one must conclude, albeit tentatively, that judges are not discriminatory in their decisions, which are primarily based on more legally relevant criteria. This does not mean, however, that discrimination does not exist in our criminal justice system. It does indicate, however, that if discriminatory decisions do occur, it happens at an earlier point in the criminal justice system. Future research on the effects of judicial characteristics should include measures of different forms of discrimination, e.g., bias due to the race of the victim and social class, in order to provide a more thorough investigation of the issue. In addition, the effects of judicial characteristics other than those included in the current analysis, e.g., political affiliation, religious preference and penal philosophy, should be examined. Qualitative data used in conjunction with empirical analyses may help shed light on the meanings attributed to particular factors by sentencing judges. Furthermore, such data would 140 provide a better understanding of the social context within which sentencing decisions are made. NOTES 1. Use of the male pronoun is not intended to ignore the increasing numbers of female judges on the American bench. Where applicable, the gender of the personal pronoun will be alternated throughout the discussion. 2. In order to demonstrate discrimination in the legal sense, the defense must show that the intent to discriminate exists. In the sentencing literature the term refers to a pattern of case outcomes that relies on a factor which is not a legitimate consideration in the sentencing decision, e.g., sex, race or religion. Thus, demonstration of intent is not required. The latter sense of the term will be employed in this paper. 3. The following figures are based on data presented in Table 1-3 on page 21 of Hagan and Bumiller (1983). 4. The following figures are based on data reported in Table 3 of Kleck (1981). 5. When a case is adjourned in contemplation of dismissal, the defendant has been adjudicated guilty but not officially convicted. Therefore, the defendant is not subject to a sentencing decision, unless rearrested on a new charge. 6. Alpert did not identify the source of the data or the statistical methods employed for this analysis. 141 ' 142 7. As in the 1979 study, Alpert neglected to fully describe the sample of individuals interviewed. 8. This is the same Metro City which provided the data base for the Kritzer and Uhlman (1977) study. 9. Cases were identified as belonging to the population of cases sentenced during 1977 through the Criminal Case Conviction Register. 10. The State Court Administrative Office was able to supply information concerning other demographic and biographic characteristics of these judges, e.g., the law school attended by the judge, his/her previous employment, the manner in which he/she assumed the bench. While these variables would have been theoretically relevant in the analysis at hand, the many instances of missing data prevented their use from being feasible. 11. The exclusion of victim race from consideration is based on problems in the coding of the data on victim characteristics. Missing cases for the race variable were included in the category "non-white". In a recent project, Bynum and his colleagues (1986) utilized presentence investigation reports in Michigan to collect the same type of data on victims as Zalman et al. (1977). Bynum reported that over 50% of the data for victim race was missing. Data for other victim characteristics had similar problems. Given that it is unlikely that 1977 presentence reports were more complete than the ones examined in 1986, it is 143 unreasonable to assume that those victims for whom race is unknown fall into the non-white category. In addition, it is impossible to identify those offenses in which the victim was not a human, e.g., burglary of a home or business. Clearly, it makes little or no sense to classify an institutional victim as either white or non-white, young or old. For these reasons, victim characteristics are not considered in this analysis, even though they may have an important effect on the sentencing decision. 12. See Hanushek and Jackson (1977) for a more technical discussion of the problems in estimating models with discrete dependent variables and the uses of PROBIT and LOGIT. 13. Several cases were coded as having received sentences that were greater than the maximum set by the legislature. These were clearly coding errors and were therefore eliminated from the sample. 14. Technically, judges are not empowered to impose a minimum sentence which is equal to the statutory maximum. According to the statute the judge must impose a minimum sentence which is no more than 2/3 of the legislatively defined maximum. However, some judges disregard this aspect of the law, so values up to 1.0 were utilized in the analysis of this variable. 15. Since the majority of non-white defendants in Detroit 144 are black, the terms "non-white" and "black" will be used interchangeably. 16. These variables were dichotomous, therefore, the mean of the variable was the proportion of the cases in the sample coded as 1. It is possible, then, to speak of the percentage of cases having a particular characteristic by multiplying the mean of the variable by 100. 17. In ordinary least squares analysis the overall fit of the model is usually determined by the amount of variance explained, R2. Although there is an R2 analogy in logit and probit models (McKelvey and Zavoina, 1975), its sampling distribution is unknown and may substantially underestimate the model's true fit. In its place, the Model Classification Table reported for each of the logistic regression tables is employed. This table estimates the overall goodness-of—fit of the model in terms of its ability to predict the outcome on the dependent variable. The percent correctly classified by the model estimates the accuracy of the prediction based upon the marginals predicted by the model. The percent correctly classified is the sum of the cells in which the predicted outcome equals the actual outcome divided by the total. The percent correctly classified by chance is based upon marginal distributions assuming the actual outcome and predicted outcomes are independent (this procedure is similar to that done for expected cell frequencies used in chi-square 145 tables). The proportion reduction in error relative to chance measures the percent of classification error by chance that is reduced by using the model. It reflects, then, the percentage of errors one would have made but no longer makes based on prediction from the model. 18. Since female judges sentenced only 9.8% of the cases in the sample, the robustness of this estimate was questionable. In addition to the possiblity that the coefficient reflected a true relationship between the judicial gender and the decision to incarcerate, there were two other possible reasons for the finding of a significant relationship. First, it is possible that variables related to the gender of the judge that were also related to the sentencing decision were omitted from the model. However, without a strong theoretical justification, this argument is weak especially given the good fit of the model to the data. Second, it is possible that this effect was due to an interaction of the gender variable with one of the other variables included in the model (Smith, 1987). In order to test this hypothesis, a series of logit equations including interactions of judicial gender with other variables in the model were estimated. In none of the equations, did the interaction variable attain significance and reduce the coefficient found for judicial gender to the point that the effect could be due to chance. In light of these analyses, we can conclude that this effect is robust. 146 19. See the discussion of judicial race in chapter 1, particularly the discussion concerning the role of black consciousness in the handling of cases. 20. The beta coefficient (B) is a standardized version of the effect of an independent variable on the dependent variable. Since the variables are standardized in the estimation of B, these coefficients may be compared within a model in order to determine the importance an effect relative to the other variables in the equation. See Hanushek and Jackson (1977) for a more technical discussion. REFERENCES REFERENCES Aberbach, J.D. and J.L. Walker. (1973) Race in the City. Boston: Little, Brown. A Alpert, L. (1981) Learning about trial judging: The socialization of state trial judges. In J.A. Cramer (ed.) Courts and Judges, pp. 105-147. Beverly Hills: Sage Publications. Alpert, L., B.M. Atkins, and R.C. Ziller. (1979) Becoming a judge: The transition from advocate to arbiter. Judicature. 62(7): 325-335. Antunes, G. and C.M. Gaitz. (1975) Ethnicity and participation: A study of Mexican-Americans, blacks and whites. American Journal 9: Sociology. 80: 1192-1211. Atkins, B.M. (1974) Opinion assignment on the United States Court of Appeals: The question of issue specialization. Western Political Quarterly. 27: 409-428. Atkins, B. and H.R. Glick (1976) Environmental and structural variables as determinants of issues in state courts of las resort. American Journal 9: Political Science. 20: 97-115. Baldus, D.C., C. Pulaski and G. Woodworth. (1983a) Comparative review of death sentences: An empirical study of the Georgia experience. Journal 9f Criminal Law and Criminology. 74: 661-753. (1983b) Discrimination and arbitrariness in Georgia's capital charging and sentencing system: A preliminary report. Unpublished manuscript, University of Iowa. (1985) Monitering and evaluating contemporary death sentencing systems: Lessons from Georgia. g.g. Davis Law Review. 18: 1327-1374. Beccaria, C. (1963) On Crimes and Punishments, trans. Henry Paolucci. New York: Bobbs-Merrill. 147 148 Becker, T. (1966) A survey study of Hawaiian judges: The effect on decisions of judicial role variations. American Political Science Review. 60: 677-680. Bell, D. (1973) Race, Racism and American Law. Boston: Little, Brown and Company. Bernard, T.J. (1983) The Consensus-Conflict Debate: Form and Content in Social Theories. New York: Columbia University Press. Bernstein, I.N.; W.R. Kelly; P.A. Doyle. (1977) Societal reaction to deviants: The case of criminal defendants. American Sociological Review. 42: 743-755. Blumstein, A.; J. Cohen; S.E. Martin; and M.H. Tonry (eds.) (1983) Research 93 Sentencing: The Search for Reform, Vol. I and II. Washington, D.C.: National Academy Press. Bowers, W.J. (1983) The pervasiveness of arbitrariness and discrimination under post-Furman capital statutes. Journal 9: Criminal Law and Criminology. 74: 1067-1100. Burns, H. (1973) Black people and the tyranny of American law. Annals g; the American Academy 9f Political and Social Science. 407: 156-166. Bynum, T.S. (1986) Michigan Corrections Project, 1986, E. Lansing, Michigan 48824. Campbell, D. and J. Feagin. (1975) Black politics in the South: A descriptive analysis. Journal 9: Politics. 37: 129-159. Chambliss, W.J. and R.B. Seidman. (1971) Law, Order and Power. Menlo Park, California: Addison-Wesley. Clemente, F. and W.J. Sauer. (1975) Voluntary association and minority status: Comment and extension. American Sociological Review. 40: 115-117. Cohen, L.E. and J.R. Kluegel. (1978) Determinants of juvenile court dispositions: Ascriptive and achieved factors in two metropolitan courts. American Sociolggical Review. 43: 162-176. Cohen, S.M. and R.B. Kapes (1978) Participation of blacks, Puerto Ricans, and whites in voluntary associations: A test of current theories. Social Forces. 56: 1053-1069. 149 Cook, B.B. (1979) Judicial policy: Change over time. American Journal pf Political Science. 23: 208-214. (1980) Political culture and selection of women judges in trial courts. In D.W. Stewart (ed.) Women ip Local Politics, pp. 46-60. London: Scarecrow Press. Crockett, G.W., Jr. (1971) Racism in the courts. Journal pf Public Law. 20: 385-389. Crockett, G.W., Jr. (1984) The role of the black judge. In Georges-Abeyie (ed.) The Criminal Justice System and Blacks, pp. 195-204. New York: Clarke Boardman. Diamond, I. (1977) Sex Roles ip the State House. New Haven, Conn.: Yale University Press. Dye, T.R. and J. Renick. (1981) Political power and city jobs: Determinants of minority employment. Social Science Quarterly. 62: 475-486. Eisenstein, J. and H. Jacob. (1977) Felony Justice: Ag Organizational Analysis pf Criminal Courts. Boston: Little, Brown. Eisinger, P.K. (1982a) Black employment in municipal jobs: The impact of black political power. American Political Science Review. 76: 380-392. (1982b) The economic conditions of black employment in municipal bureaucracies. American Journal pf Political Science. 26: 754-771. Engle, C. D. (1971) Criminal Justice in the City: Stud y of Sentence Severity and Variation— in the Philadelphia Criminal Court System. Unpublished PhD. dissertation, Temple University. Epstein, g.§, (1970 Woman's Place: Options and Limits in Professional Careers. Berkeley, California: University of California Press. Erikson, R. and N. Luttberg (1973) American Public Opinion: Its Origins, Content and Impact. New York: Wiley. Feagin, J.R. (1970) Black elected officieals in the south: An exploratory analysis. In J.R. van Der Slik (ed.) Black Conflict with White America, pp. 107-122. Columbus, Ohio: Charles E. Merrill Publishing Co. Feinman, C. (1986) Women ip the Criminal Justice System. New York: Praeger. 150 Frank, J. (1949) Facts are guesses. In J. Frank (ed.) Courts pp Trial: Myth Egg Reality 1p American Justice. Princeton: Princeton University Press. Reprinted in S. Goldman and A. Sarat (eds.) American Court Systems: Readings 1p Judicial Process and Behavior. San Francisco: W.H. Freeman and Company (1978). Frazier, C.E. and E.W. Bock. (1982) Effects of court officials on sentence severity. Criminology. 20(2): 257-272. Garber, S.; S. Klepper; and D. Nagin. (1983) The role of extra legal factors in determining criminal case disposition. In A. Blumstein, J. Cohen, S.E. Martin and M.H. Tonry (eds.) Research pp Sentencing: Egg Search for Reform, 291. ‘II, pp. 129-183. Washington, D.C.: National Academy. Gaylin, W. (1974) Partial Justice. New York: Alfred A. Knopf. Georges-Abeyie, D. (1984) The criminal justice and minorities--a review of the literature. In D. Georges-Abeyie (ed.) Criminal Justice System and Blacks, pp. 125-158. New York: Clarke Boardman Company, Ltd. Glueck, S. and E. Glueck. (1964) Ventures 1p Criminology. Cambridge, Mass.: Harvard University Press. Gibson, J.L. (1977) Discriminant functions, role orientations and judicial behavior: Theoretical and methodological linkages. Journal pf Politics. 39: 984-1007. (1980) Environmental constraints on the behavior of judges: a representational model of judicial decisionmaking. Law & Society Review. 14: 343-370. (1983) From simplicity to complexity: The development of theory in the study of judicial behavior. Political Behavior. 5: 911-924. ' (1981) Personality and elite political behavior: The influence of self esteem on judicial decision making. Journal pf Politics. 43: 104-125. (1978) Race as a determininant of criminal sentences: A methodological critique and a case study. Law & Society Review. 12:455-478. Goldman, S. (1979) The effect of past judicial behavior on subsequent decision making. Jurimetrics Journal. 19: 208-217. 151 (1975) Voting behavior on the U.S. circuit of appeals revisited. American Political Science Review. 69: 491-506. Green, E. (1961) Judicial Attitudes 1p Sentencing. Westport, Conn.: Greenwood Press. Greenberg, D. (1981) Crime and Capitalism. Palo Alto, CA: Mayfield Publishing. Greenberg, 5.3. (1974) Politics and Poverty. New York: John Wiley. Gross, S.R. and R. Mauro. (1984) Patterns of death: An analysis of racial disparities in capital sentencing and homicide victimization. Stanford Law Review 37: 27-153. Grossman, J.B. (1967) Social backgrounds and judicial decisions: Notes for a theory. Journal pf Politics. 29: 334-351. Gruhl, J.; C. Spohn and S. Welch. (1981) Women as policymakers: The case of trial judges. American Journal pf Political Science. 25(2): 308-322. Hagan, J. (1974a) Extra-legal attributes and criminal sentencing: An assessment of a sociological viewpoint. Law & Society Review. 8: 357-384. (1974b) Parameters of criminal prosecution: an application of path analysis to a problem of criminal justice. Journal pf Criminal Law and Criminology. 65(4): 536-544. Hagan, J. and I.N. Bernstein. (1979) Conflict in context: The sanctioning of draft resisters, 1963-76. Social Problems. 27(1): 109-122. Hagan, J. and K. Bumiller. (1983) Making sense of sentencing: A review and critique of sentencing research. In A. Blumstein, J. Cohen, S.E. Martin, and M.H. Tonry (eds.) Research pp Sentencing: The Search for Reform, pp. 1-54. Washington, D.C.: National Academy Press. Hanushek and Jackson (1977) Statistical Methods for Social Scientists. Orlando: Academic Press, Inc. Hogarth, J. (1971) Sentencing as a Human Process. Toronto: University of Toronto Press. Horowitz, D. (1977) The Courts and Social Policy. Englewood Cliffs, N.J.: Prentice-Hall. 152 Howard, J. W. (1981) Courts of Appeals Ln the Federal Judicial System: A Study of the Second, Fifth, and District pf Columbia Circuits. Princeton: Princeton University Press. Jaros, D. and R. Mendelsohn (1967) The judicial role and sentencing behavior. Midwest Journal pf Political Science. 11: 471-488. Karnig, A.K. and S. Welch. (1980) Black Representation and Urban Policy. Chicago: University of Chicago Press. Kleck, G. (1981) Racial discrimination in criminal sentencing: A critical evaluation of the evidence with additional evidence on the death penalty. American Sociological Review. 46: 783-805. Klepper, S.; D. Nagin and L. Tierney (1983) Discrimination in the criminal justice system: A critical appraisal of the literature. In Blumstein, A., J. Cohen, S. E. Martin, and M. H. Tonry (eds. ) Research on Sentencing: The Search for Reform, pp. 55- 128. Washington, D. C.: National Academy Press. Klobus-Edwards, P.; J.N. Edwards; and D.J. Klemmack. (1978) Differences in social participation: Blacks and whites. Social Forces. 56: 1035-1052. Kritzer, H.M. (1979) Federal judges and their political environments: The influence of public opinion. American Journal pf Political Science. 23: 194-207. (1978) Political correlates of the behavior of federal district judges: A "best case" analysis. Journal pf Politics. 40: 25-58. Kritzer, H.M. and T.M. Uhlman. (1977) Sisterhood in the courtroom: Sex of judge and defendant in criminal case dispositon. The Social Science Journal. 14(2): 77-88. Lamb, C. (1976) Exploring the conservatism of federal appeals court judges. Indiana Law Journal. 51: 257-279. (1975) Political Power ip Poor Neighborhoods. New York: John-Wiley. Lefcourt, R. (1974) Law against the people. In Richard Quinney (ed.) Criminal Justice ip America: A Critical Understanding. Boston: Little, Brown and Co. Levin, M. (1977) Urban Politics and the Criminal Courts. Englewood Cliffs, N.J.: Prentice-Hall. 153 London, B. (1975) Racial differences in social and political participation: It's not simply a matter of black and white. Social Issues Quarterly. 56: 274-286. Long, E.; J. Long; W. Leon; and P2 Weston. (1975) American Minorities: The Justice Issue. Englewood Cliffs, N.J.: Prentice-Hall. McPherson, J.M. (1971) Correlates of social participation: A comparison of the ethnic community and compensatory theories. Sociological Quarterly. 18: 209-222. Meier, K.J. and R.B. England. (1984) Black representation and educational policy: Are they related? American Political Science Review. 78: 392-403. Michigan Department of Corrections (1986) 1985 Annual Report. Lansing, Michigan. Milbrath, L.W. and M.L. Goel. (1977) Political Participation. Chicago: Rand McNally. Myers, M.A. (1987) Economic inequality and discrimination in sentencing. Social Forces. 65(3): 746-766. Nagel, 5.8. (1962) Ethnic affiliations and judicial propensities. Journal pf Politics. 24: 92-110. Nelson, W.E. and Meranto, P.J. (1977) Electing Black Mayors: Political Action 13 the Black Community. Columbus: Ohio State University. Parsons, T. and R.F. Bales. (1955) Family, Socialization and Interaction Process. New York: Free Press of Glencoe. Partridge, A. and W.B. Eldridge. (1974) The Second Circuit Sentencing Study: A Report pp the Judges 9; the Second Circuit. Washington, D.C.: Federal Judicial Center. Peterson, R.D. and J. Hagan (1984) Changing conceptions of race: Towards an account of anomalous findings of sentencing research. American Sociological Review. 49: 56-70. Quinney, R. (1975) Criminolggy: Analysis and Critique pf Crime 1p America. Boston: Little, Brown and Company. Radelet, M.R. and G.L. Pierce (1985) Race and prosecutorial discretion in homicide cases. Law and Society Review. 19(4): 587-621. Schmidhauser, J. (1979) Judges and Justices: The Federal Appellate Judiciary. Boston: Little, Brown. 154 (1960) The Supreme Court: Its Politics, Personalities and Procedures. New York: Holt, Rinehart and Winston. Schubert, G. (1980) The Judicial Mind Revisited: Psychometric Analysis pf Supreme Court Ideology. New York: Oxford University Press. Shingles, R.D. (1981) Black consciousness and pollitical participation: The missing link. American Political Science Review 75: 76-91. Smith, D.A. (1987) information provided in a conversation. Soule, J. and W. McGrath. (1977) A comparative study of male-female political attitudes at citizen and elite levels. In M. Githens and J. Prestage (eds.) A Portrait pf Marginality. New York: McKay. Steggert, F.T. (1975) Community Action Groups and City Government. Cambridge, Mass.: Ballinger. Thomson, R.J. and M.T. Zingraff (1981) Detecting sentencing disparity: some problems and evidence. American Journal pf Sociology. 86:869-880. Turk, A. (1969) Criminality and Legal Order. Chicago: Rand McNally. . Uhlman, T.M. (1979) Racial Justice. Lexington, Mass.: Lexington Books. Ulmer, 8.5. (1974) Dimensionality and change in judicial behavior. In J.F. Herndon and J.L. Bernd (eds.)Mathematical Applications ip Political Science--!;;. Charlottesville: University Press of Virginia. Unnever, J.D. (1982) Direct and organizational discrimination in the sentencing of drug offenders. Social Problems. 30: 212-225. Van Koppin, P.J. and J. Ten Kate. (1984) Individual differences in judicial behavior: personal characteristics and private law decision-making. Law & Society Review, 16(2): 225-247. Verba, S. and N.H. Nie. (1972) Participation 12 America. New York: Harper and Row. Vold, G.B. and T.J. Bernard. (1986) Theoretical Criminology, 3rd ed. New York: Oxford University Press. 155 Vorenberg, J. (1981) Decent restraint of prosecutorial power. Harvard Law Review. 94: 1521-1573. Welch, S.; J. Gruhl; and C. Spohn (1984) Sentencing: the influence of alternative measures of prior record. Criminology. 22(2): 215-227. Welch, S. and A.K. Karnig. (1979) The impact of black elected officials on urban social expenditures. Policy Studies Journal. 7: 380-392. Wellford, C.F. (1975) Labelling theory and criminology: An assessment. Social Problems. 22: 332-345. Wilkins, L.T.; J.M. Kress; D.M. Gottfredson; J.C. Calpin; and A.M. Gelman. (1976) Sentencing Guidelines: Structuring Judicial Discretion. A report of the Criminal Jusitice Research Center. One Alton Road. Albany, N.Y. 12203. Williams, J.A.; N. Babchuck; and D. Johnson. (1973) Voluntary associations and minority status: A comparative analysis of anglo, black and Mexican Americans. American Sociological Review. 38: 637-646. Wilson, J.Q. (1975) Thinking About Crime. New York: Basic Books. , Wright, B.M. (1984) A view from the bench. In D. Georges-Abeyie (ed.) The Criminal Justice System and Blacks, pp. 205-218. New York: Clarke Boardman. Ziller, R.C.; W.F. Stone; R.M. Jackson; and N.J. Terbovic. (1979) Self-other orientations and political behavior. In M.G. Hermann, with T.W. Milburn (eds.) Psychological Examination pf Political Leaders, pp. 335-369. New York: Free Press. Cases Cited Bppyp y. the Board pf Education, 347 S. Ct. 43 (1975) Green y. U.S., 356 U.S. 165 (1958) ngpy y, Spgpg, 328 A 2d 293 (Ct. App. 1974) Spgpe y. Sharpe, 541 P 2d 41 (1975) HICHIGRN STATE UNIV. I L B 312930152728 RRRIES IIIHHHI 79