INTRODUCTION BACKGROUND RESEARCH QUESTION

Transcription

1 Disparity under Structured Sentencing in North Carolina: Do similarly situated offenders receive different outcomes based on legally irrelevant factors? by Michelle L. Hall A paper submitted to the faculty of the University of North Carolina at Chapel Hill in partial fulfillment of the requirements for the degree of Master of Public Administration Spring 2011 The attached paper represents work done by a UNC-Chapel Hill Master of Public Administration student. It is not a formal report of the School of Government, nor is it the work of School of Government faculty. Executive Summary Structured sentencing in North Carolina was implemented in 1994 to create a predictable, stable, and equitable sentencing system. This study is modeled after a 2002 disparity study conducted by the North Carolina Sentencing and Policy Advisory Commission, and seeks to understand whether similarly situated offenders under structured sentencing receive breaks (charge reductions, non-active sentences, and minimum sentence lengths) based upon legal and extralegal factors. Statistical analysis reveals that while outcomes for offenders can be predicted by legally relevant factors (e.g., offense seriousness and criminal history), legally irrelevant factors (such as age and gender) also influence outcomes. The findings are consistent with the 2002 study and show that male offenders are less likely to get breaks than female offenders, offenders with privately retained attorneys are more likely to get better results than those with private assigned counsel or public defenders, older offenders receive certain breaks, sentencing outcomes vary across judicial divisions, and offenders who opt for a jury trial receive worse outcomes. Race was found to be significant in the current study, but not in the previous study.

2 INTRODUCTION Prior to 1994, North Carolina adhered to a sentencing system known as fair sentencing, which gave judges wide discretion to determine criminal sentences. 1 In 1994, the North Carolina General Assembly enacted structured sentencing in North Carolina. The intent of structured sentencing was to establish fair, rational and consistent sentencing policies and construct a system in which these policies are balanced with available correctional resources. 2 This new system mandated the use of a sentencing grid, which defined minimum and maximum sentence ranges based on offenders prior criminal history and current conviction class, 3 and was designed to result in predictable and equitable sentencing outcomes for offenders. The North Carolina Sentencing and Policy Advisory Commission (Sentencing Commission) conducted the first review of sentencing practices under structured sentencing system in The study focused on two remaining areas of discretion in North Carolina s criminal justice system: the plea negotiation and sentence disposition processes. It examined both legal and extralegal factors affecting breaks 4 for offenders at those two stages. The findings revealed certain extralegal factors (age, sex, judicial division of conviction, defense attorney type, and mode of disposition) were statistically significant in predicting breaks for offenders, suggesting disparate treatment within the criminal justice system. 5 This study is modeled after the 2002 study. Using more recent data, it attempts to determine if and to what extent legal and extralegal factors influence sentencing outcomes and then compares the findings to the 2002 study. BACKGROUND Twenty states and the District of Columbia have some form of sentencing guidelines. 6 These guidelines are designed to provide structure at the criminal sentencing stage by specifically defining offense and offender elements that should be considered in each case, 7 thus promoting similar sentencing outcomes for similarly situated offenders (i.e., those who have committed similar offenses and have similar criminal histories). 8 By limiting judicial discretion these guidelines seek to eliminate the possibility of disparity in the criminal justice system. Whether structured sentencing systems have eliminated disparity, however, has been the subject of much debate. 9 Several experts have observed sentencing guidelines, while limiting judicial discretion, shift discretion to prosecutors. 10 This prosecutorial discretion is perhaps greatest in the plea bargaining stage, where prosecutors are given wide latitude to accept and reject plea bargains with limited oversight from the courts. 11 Research regarding prosecutorial and judicial discretion under sentencing guidelines has resulted in mixed conclusions. Several studies found extralegal factors (race, gender, socio-economic status, etc.) do influence sentencing outcomes with minorities, poor, and male offenders receiving harsher outcomes. 12 Other studies have found the opposite. 13 The 2002 study concluded, What is of possible concern is the apparent influence of extralegal factors such as age, sex, judicial division of conviction, defense attorney type and mode of disposition [in producing favorable decisions for offenders]. 14 This suggests a need to revisit whether structured sentencing in North Carolina is achieving its goal of consistent sentencing practices. RESEARCH QUESTION Do legally irrelevant factors (i.e., those unrelated to an offender s criminal history or offense characteristics) affect sentencing outcomes for similarly situated offenders in North Carolina? 1

3 METHODOLOGY AND DATA DESCRIPTION Court data from for all felony charges resulting in felony or misdemeanor convictions were collected from the Sentencing Commission. 15 To determine which legal (e.g., offense class, offense type, criminal history) and extralegal variables (e.g., race, gender, age, etc.) influence the probability in each model, binary logistic regressions were conducted on Models 1-3 and a linear regression was conducted on Model 4 (outlined in Table 1, below; see Appendix B for full variable list with definitions). Four different areas of judicial or prosecutorial discretion were analyzed. Table 1 Model Dependent Variable 16 Data Description N Model 1 Probability that offender received reduction Felony charges resulting in felony or 46,259 from felony charge to misdemeanor conviction misdemeanor convictions Model 2 Probability that offender received reduction from felony charge to lesser felony charge of at least one charge class Felony charges resulting in felony convictions 32,956 Model 3 Probability that offender received active sentence when one was not mandated Felony convictions involving judicial discretion to impose active or inactive sentences Model 4 Length of minimum sentence imposed Felony convictions for which active sentences were imposed Descriptive Statistics All Felony Charges Year (N=46,259) See also Appendix C for additional descriptive statistics 21,182 12,177 Average Percent Poverty in District Mean Median Standard Deviation 11 4 Range White (43%) Non White (57%) Privately Retained Attorney (20.2%) Private Assigned Counsel (51.3%) Public Defender (24.3%) Waived (4.2%) Male (82.1%) Female (17.9%) This study focuses only on discretionary decisions made during the conviction and sentencing stages of the criminal justice process. It does not examine arrest data or original charges that differ from indicted charges, nor does it attempt to determine whether disparate treatment occurs at those stages in the criminal justice process. 2

4 FINDINGS Table 2: Regression Results: Independent Variable Influence on Probability of Outcomes (Models 1-4) Only statistically significant results show; not all models included all independent variables. Independent Variable Influence Models with different dependent variables + = increases probability of outcome - = decreases probability 1: Misdemeanor conviction 2: Felony charge reduction 3: Active sentence when judge has discretion Variables (compared to reference category) Charge Class (Models 1&2) OR Conviction Class (Models 3&4) Charge Type (Models 1&2) OR Conviction Type (Models 3&4) Prior Record Priors=at least one prior conviction 4: Severity of minimum sentence length imposed A - + B B C D E F G H - (reference category) + + I (reference category) - (reference category) (reference category) Drug Person Property (reference (reference category) (reference category) (reference category) category) Other - - Priors - Prior I (reference category) (reference category) (reference category) Prior II Prior III + + Prior IV Prior V Prior VI Charge Reduction + + Extralegal Variables (compared to reference category) Offender - (older less likely to get break) Jurisdiction (Male) Race (Nonwhite) Judicial Division - (older less likely to get break) - (older more likely to get break) or depending on Division + or depending on Division + - (older more likely to get break) - (compared to Division 1) - (compared to Division 1) Percent + - Poverty Other Jury Trial Privately Retained Attorney Court + (Superior Court) 3

5 Finding 1: and extralegal variables are significant predictors of the probability offenders receive breaks. Variables In all models, not surprisingly, legal variables influenced the probability offenders received breaks (see Table 2). Offenders charged with more serious felony offenses were less likely to receive misdemeanor convictions. Offenders charged with more serious offenses were more likely to receive felony charge reductions of at least one offense class. This is likely due to the fact that plea bargains usually move charges downward on the sentencing grid. More extensive criminal history decreased the likelihood offenders received misdemeanor convictions and felony charge reductions, increased the likelihood offenders received an active sentence, and increased the severity of the minimum sentence imposed. These findings indicate judges do take into account the legal considerations dictated by the sentencing grid. Extralegal Variables Extralegal variables also influenced whether offenders received breaks. On the whole, male offenders received harsher outcomes compared to female offenders. Males were less likely to receive misdemeanor convictions from felony charges, less likely to receive felony charge reductions, and more likely to be given active sentences. This gender disparity may be due to other factors (not controlled for in this study), for example, whether or not female offenders are primary caretakers. Offenders who retained private attorneys received more favorable outcomes at all discretionary points, compared to those represented by a public defender (PD) or private assigned counsel (PAC). Offenders with retained attorneys were 10% more likely to receive misdemeanor convictions and 12% less likely to receive active sentences, compared to indigent offenders. Judicial divisions were significant in predicting sentencing outcomes. In some divisions, offenders were more likely to receive misdemeanor convictions, felony charge reductions, and non-active sentences, while in other divisions, offenders were less likely to receive those breaks. This variation among judicial divisions may be because North Carolina elects judges and district attorneys, and those who get elected could reasonably be assumed to reflect the values of the electorate. For example, some localities may, on the whole, favor rehabilitative modes of punishment (generally, probation) over retributive modes (imprisonment), while other localities may feel the opposite. Whatever the precise reasons, collectively these differences have a measureable impact on plea bargaining and sentencing practices across the state. Younger offenders tended to receive breaks at the plea bargaining stage (they were more likely to receive misdemeanor convictions and felony charge reductions) but older offenders tended to receive breaks at the sentencing stage (older offenders were less likely to receive active sentences and received less severe minimum sentences). Race was significant with regard to imprisonment; nonwhite offenders were more likely to receive active sentences compared to white offenders. Offenders who opted for jury trials fared worse than those who pleaded guilty. While only about 1.6% of cases were tried by juries in , those offenders were less likely to receive breaks compared to those who pleaded guilty. Finding 2: Significant variables are generally consistent with the 2002 Study. 4

6 Unsurprisingly, both studies found offense seriousness and criminal history to be significant in predicting whether offenders receive breaks. Offenders with more serious crimes and more extensive criminal history tended to receive less favorable outcomes. This suggests that judges and prosecutors continue to take those factors into account as they must under the law. Extralegal variables constant between the two studies include age, judicial division, type of attorney, trial by jury, and gender. This suggests that extralegal factors influencing plea bargaining and sentencing practices in North Carolina have remained the same since 2002 with some offenders receiving worse outcomes. Overall, the consistency of the findings in the 2002 study and the current study suggests that these biases may be inherent in the system, but they are becoming neither more or less pronounced over time. On the other hand, race was found to be significant in the current study, but not in the previous study. While race was found to be a significant variable only in the discretionary decision of imposing an active sentence, its presence as a significant variable at all is disconcerting, suggesting the influence of this variable is worsening compared to the 2002 findings. RECOMMENDATIONS Extralegal variables found to influence sentencing outcomes merit further examination. Regular disparity studies are necessary to identify recurring patterns of sentencing disparity in North Carolina. Though race was not found to be statistically significant in the 2002 study, the fact that it was found to be significant in one of the four models in this study merits particular examination. To even begin to address disparity, policymakers and practitioners must know if and where disparities exist to determine appropriate solutions. Future studies might, for instance, seek to control for additional variables, such as employment status, and education level, to refine understanding of how extralegal variables influence sentencing outcomes. Consider whether further limiting discretion would achieve more consistent results. Policymakers may consider the tension between discretion and consistency to determine if the current sentencing grid achieves the proper balance. The grid could be modified to narrow the range of permissible sentence lengths in a particular cell on the grid or to reduce the number of cells in which judges have a choice between prison and probation. However, this would essentially eliminate the need for judges to determine and tailor individual sentences. Arguably, this would also shift discretion entirely to the prosecutors at the plea bargaining stage. Prosecutorial guidelines (regulation of prosecutorial practices) have been attempted with limited success; however, research suggests they may be successful when combined with sentencing guidelines. 17 Understanding the tradeoffs and impacts of furthering limiting discretion is critical to any decisions or policy changes attempting to address disparity. Further research into the effectiveness of PDs and PACs is needed. This study and the 2002 study found representation by a retained private attorney resulted in more favorable outcomes for offenders across all discretionary decision points (compared to PACs and PDs). The constitutional right to retain a lawyer cannot be limited. However, the system for training and appointing counsel for indigent offenders should be made as effective as possible. The effectiveness of PDs compared to PACs could not be determined in this study and merits further attention. 5

7 1 Freeman, Lorrin. The North Carolina Sentencing and Policy Advisory Commission: A History Of Its Creation And Its Development Of Structured Sentencing. North Carolina Sentencing and Policy Advisory Commission (Updated August 2009). 2 North Carolina Sentencing and Policy Advisory Commission. Sentencing Practices Under North Carolina s Structure Sentencing Laws. March See supra at note 1. 4 A break is defined as the following: (1) a reduction from felony charge to misdemeanor; (2) a reduction from a felony charge to a lesser felony charge of at least one charge class; (3) imposition of an inactive sentence when an active sentence is not mandatory; and (4) imposition of the minimum sentence length. 5 See supra at note 2. 6 Kauder, Neal. Ostrom, Brian. State Sentencing Guidelines: Profiles and Continuum. National Center for State Courts Id. 8 Id. 9 Hagan 1974; Albonetti 1997; Alschuler 1978; Miethe 1987; Standen 1993; Tonry 1996; Wooldredge Ball 2006; Wilmot 2004; Standen 1993; Miethe, 1988; Alschuler Standen 1993; Wooldredge Piquero 2004; Albonetti 1997; Ulmer and Kramer, Ball 2006; Hagan, See supra at note See Appendix for descriptive statistics of dataset. 16 The areas of discretion (i.e. the dependent variables) are the same as those used in the 2002 Sentencing Commission study to allow for direct comparison of findings. 17 Wright, Ronald F. Sentencing Commissions as Provocateurs of Prosecutorial Self-Regulation. Columbia Law Review 105 (4)

8 Acknowledgements I would like to thank my Committee members Maureen Berner (Chair), Michael Smith, and James Markham for their input and guidance on this paper. Their statistical, legal and practical expertise were essential to the success of my research. 7

9 APPENDIX A Works Cited Albonetti, Celesta A. Sentencing Under the Federal Sentencing Guidelines: Effects of Defendant Characteristics, Guilty Pleas, and Departures on Sentence Outcomes for Drug Offenses, Law & Society Review 31(4): Alschuler, Albert. Sentencing Reform and Prosecutorial Power. University of Pennsylvania Law Review 126: Ball, Jeremy D. Is It a Prosecutor s World?: Determinants of Count Bargaining Decisions. Journal of Contemporary Criminal Justice 22: Coffee, John C. Jr., and Michael Tonry. Hard Choices: Critical Trade-offs in the Implementation of Sentencing Reform through Guidelines. Pp in Michael Tonry and Franklin E. Zimring (eds) Reform and Punishment: Essays on Criminal Sentencing. Chicago: University of Chicago Press Engen, Rodney L. and Randy Gainey. Modeling the Effects of and Extra-legal Factors Under Sentencing Guidelines: The Rules Have Changed. Criminology 38(4): Frase, Richard S. State Sentencing Guidelines: Diversity, Consensus, and Unresolved Policy Issues. Columbia Law Review Vol.105 No. 4: Freeman, Lorrin. The North Carolina Sentencing and Policy Advisory Commission: A History Of Its Creation And Its Development Of Structured Sentencing. North Carolina Sentencing and Policy Advisory Commission (Updated August 2009). Gelacak, Michael S., Ilene H. Nagel and Barry L. Johnson. Departures Under the Federal Sentencing Guidelines: An Empirical and Jurisprudential Analysis. Minnesota Law Review 81: Hagan, John. Extra-legal Attributes and Criminal Sentencing: An Assessment of a Sociological Viewpoint. Law and Society Review 8: Kauder, Neal B. and Ostrom, Brian J. State Sentencing Guidelines: Profiles and Continuum. National Center for State Courts. July Kramer, John and Jeffery T. Ulmer. Sentencing Disparity and Departure from Guidelines. Justice Quarterly 13: Miethe, Terence D. Charging and Plea Bargaining Under Determinate Sentencing: An Investigation of the Hydraulic Displacement of Discretion. Journal of Criminal Law and Criminology 78(1):

10 Nagel, Ilene H. and Stephen J. Schulhofer. A Tale of Three Cities: An Empirical Study of Charging and Bargaining Practices Under the Federal Sentencing Guidelines. Southern California Law Review 66: North Carolina Sentencing and Policy Advisory Commission. A Citizen s Guide to Structured Sentencing. Updated August North Carolina Sentencing and Policy Advisory Commission. Sentencing Practices Under North Carolina s Structure Sentencing Laws. Report to the North Carolina General Assembly. March Piquero, Nicole Leeper. Davis, Jason L. Extralegal Factors and the Sentencing of Organizational Defendants: An examination of the Federal Sentencing Guidelines. Journal of Criminal Justice 32: Standen, Jeffrey. Plea Bargaining in the Shadow of the Guidelines. California Law Review 81(6): Steffensmeier, Darrell, Jeffery Ulmer, and John Kramer. The Interaction of Race,, and in Criminal Sentencing: The Punishment Cost of Being Young, Black, and Male. Criminology 36(4): Tonry, Michael. Sentencing Matters. New York: Oxford University Press Wilmot, Keith A. Spohn, Cassia. Prosecutorial Discretion and Real-Offense Sentencing: An Analysis of Relevant Conduct Under the Federal Sentencing Guidelines. Criminal Justice Policy Review 15(3): Wooldredge, Timothy G., Rauschenberg, Fritz. (Un)anticipated Effects of Sentencing Reform on the Disparate Treatment of Defendants. Law & Society Review 39(4):

11 APPENDIX B DEFINITIONS To allow for comparison, definitions were modeled after those used in the 2002 Sentencing Commission disparity study. General Terms Disparity or ly Relevant Extralegal or ly Irrelevant Structured Sentencing Sentencing Guidelines (Presumptive) Sentencing grid Cell Discretion Charge Private Attorney Variable or Factor Active Sentence Non-active Sentence Prior Record Level Break Operational Definition Unfavorable decisions in a case where extralegal factors are used to arrive at those decisions 1 Factors a judge can consider in sentencing, as defined by the Structured Sentencing Act: offense class/type and prior record Factors unrelated to offense class/type and prior record 2 Method of sentencing offenders in North Carolina; classifies offenders on basis of severity of crime and extent of their prior criminal record 3 Range of penalties for an offense based on the seriousness of the offense and the offender s criminal history; judges are expected to adhere to ranges and must document any departures 4 Felony or misdemeanor punishment chart that defines sentencing ranges based on offense class and prior record level Specific place on the sentencing grid Use of judgment to arrive at decisions in the criminal justice process that are not predetermined Refers to the most serious charge in a charging event Privately retained counsel (not publicly appointed attorneys or public defenders) Descriptor defined as either legal or extralegal used to assess impact related to offender outcomes Offender is sentenced to prison Offender is not sentenced to prison Six levels representing the extent and gravity of an offender s criminal record; an offender with little or no prior record falls into Level I; an offender with a very violent or extensive prior record falls into Level VI 5 A discretionary charging or sentencing decision that results in a more favorable outcome for an offender (e.g., a charge reduction) 6 1 Definition of disparity in the criminal justice process used in Sentencing Practices Under North Carolina s Structured Sentencing Laws. North Carolina Sentencing and Policy Advisory Commission Similar definition also used in Wooldredge, 2005; Albonetti, 1997; Standen, 1993; Miethe, Definition based on one used in Sentencing Practices Under North Carolina s Structured Sentencing Laws. North Carolina Sentencing and Policy Advisory Commission Similar definition also used in Piquero, 2004; Engen, 2000; Hagan, Definition of structured sentencing defined in A Citizen s Guide to Structured Sentencing. North Carolina Sentencing and Policy Advisory Commission. Updated August Definition of sentencing guidelines as defined by Federal Sentencing Guidelines: Background, Analysis, and Policy Options. Congressional Research Service. Report to Congress. Updated July Definition of prior record level defined in A Citizen s Guide to Structured Sentencing. North Carolina Sentencing and Policy Advisory Commission. Updated August

12 Variables 7 Operational Definition Offense type Offenses classified as either person, property, drug, or other Offense seriousness Offenses that are higher on the sentencing grid are more serious, with Class A being the most serious and Class I being the least serious (comparatively) Offense class Defined by the sentencing grid, categorized by letters A-I Charge reduction Whether an offender received a reduction from a felony charge to a misdemeanor conviction or to a lesser felony charge of at least one charge class Criminal history Existence of prior criminal record including prior convictions (see also Prior Record Level ) Priors Existence of at least one prior conviction 8 Extralegal Variables Operational Definition at sentencing Male or female Race Defined as white or non-white Judicial Division Judicial Division of conviction (based on Superior Court districts) Percent Poverty Average percent of individuals living below the poverty level within an offender s prosecutorial district; using 2009 Census data, average percent poverty is calculated using the mean of the average poverty levels of each county within districts Type of Attorney Retained private attorney, public defender, or private assigned counsel Mode of Disposition Plea bargain or jury trial Court Superior Court, District Court, or H & I Felony District Court 6 Definition of break used in Sentencing Practices Under North Carolina s Structured Sentencing Laws. North Carolina Sentencing and Policy Advisory Commission variables and extralegal variables (where variables are the same) use the same operational definitions as Sentencing Practices Under North Carolina s Structured Sentencing Laws. North Carolina Sentencing and Policy Advisory Commission Prior record levels for felony and misdemeanor convictions are calculated differently. For Model 1, where felony charges and misdemeanor convictions are combined, the variable Priors is used as a substitute for Prior Record Level to account for the different calculations, while still allowing criminal history to be accounted for. 11

13 APPENDIX C DESCRIPTIVE STATISTICS Felony Charges Offenders Charged with Felonies by The percentage of males charged with felonies in outnumbered the percentage of females. In each of the four models, which represent progressive stages in the charging and sentencing process, the percentage of males increased (i.e., in Model 4, 92.4% of offenders were male versus Model 1 where 82% were male) Male (82.1%) Female (17.9%) Offenders Charged with Felonies by Race (White versus Non-white) Overrepresentation of minority populations (by race and ethnicity) in the criminal justice system is outside the scope of this study. However, it is important to note the felony charge population makeup by race (defined in this study as white or non-white ), as race was found to be statistically significant in two of the four models. White (43%) Non White (57%) 12

14 Type of Charged Offense In , Property Charges were the most common, followed by drug charges Drug Charge (29.1%) Property Charge (42.6%) Person Charge (14.8%) "Other" Charge (11.2%) Prior Convictions/Prior Criminal History Most offenders charged in had at least one prior conviction. New offenders made up only one-fifth of all felony charges Prior Record (78.2%) No Prior Record (21.8%) 13

15 Jury Trial versus Plea Bargain An overwhelming majority of cases are disposed through the criminal justice process by plea bargain. Only a tiny percent of offenders were convicted by a jury. Jury Trial (1.6%) Pleaded guilty (98.4%) Type of Attorney Most offenders are determined to be indigent and are therefore eligible for representation by either a public defender or publicly appointed counsel (both funded through the State). Only a small percentage chose to waive representation. Privately Retained Attorney (20.2%) Private Assigned Counsel (51.3%) Public Defender (24.3%) Waived (4.2%) 14

16 APPENDIX D REGRESSION RESULTS MODEL 1: Probability of Receiving a Misdemeanor Conviction from a Felony Charge Effects of and Extralegal Factors N=46,259 Felony charges resulting in either felony or misdemeanor convictions ( ) (Overall probability of receiving a misdemeanor conviction from a felony charge=29.3%) Variables Included in Model 1* Reference Category (where applicable) Charge Class (A-H) Class I Charge Charge Type Property Charge Priors Extralegal Percent Poverty Judicial Division Division 1 Jury Trial Male Race Non-white Court Data Source: NC Sentencing and Policy Advisory Commission 2008/2009 Felony and Misdemeanor convictions *Please see Appendix B for Variable Definitions Significant Variables in Model 1 (p=.05) Extra Charge Class Charge Type Prior Record Judicial Division (3-8) Percent Poverty Significant Variables in Model 1 that DECREASE the likelihood of a misdemeanor conviction from a felony charge (compared to reference category) Extra Charge Class B1-H (Compared to Class I) Charge Type Drug or Other Judicial Division 3-8 (Compared to Property Charge) (Compared to Division 1) Priors (Compared to no prior record) Significant Variables in Model 1 that INCREASE likelihood of a misdemeanor conviction from a felony charge (compared to reference category) Extra Charge Type, Person (Compared to Property Charge) Percent Poverty in Prosecutorial District Judicial Division 2 (Compared to Division 1) 15

17 MODEL 2: Probability of Receiving a Misdemeanor Conviction from a Felony Charge Effects of and Extralegal Factors N=32,956 Felony charges resulting in felony convictions ( ) Overall probability of receiving a felony charge reduction of at least one class=21.5% Variables Included in Model 2 Reference Category (where applicable) Charge Class (A-I; Charge H excluded) Class H Charge Charge Type Property Charge Prior Record Levels II-VI Prior Record Level I Extralegal Percent Poverty Judicial Division Division 1 Jury Trial Male Race Non-white Court Significant Variables in Model 2 (p=.05) Extra Charge Class Charge Type Prior Record Judicial Division Percent Poverty Court Significant Variables in Model 2 that DECREASE the likelihood of a felony charge reduction of at least one class (compared to reference category) Extra Charge Class A-G (Compared to Class H) Charge Type Drug Trafficking or Other (Compared to Property Charge) Prior Record (Compared to Level I) Judicial Division 2-8 (Compared to Division 1) Jury Trial Superior Court (Compared to H&I Felony District Court) Significant Variables in Model 2 that INCREASE likelihood of a felony charge reduction of at least one class (compared to the reference category) Extra Charge Type, Drug (Compared to Property Charge) Percent Poverty in Prosecutorial District 16

18 MODEL 3: Probability of Receiving an Active Sentence when an Active Sentence is Non- Mandatory Effects of and Extralegal Factors N=21,182 Felony convictions that fell into discretionary cells on sentencing grid (2008/2009) Overall probability of receiving an active sentence when judge has discretion=37.9% Variables Included in Model 3 Reference Category (where applicable) Conviction Class (E-H) 9 Class I Conviction Conviction Type Property Conviction Prior Record Level (II-VI) Prior Record Level I Charge Reduction Extralegal Percent Poverty Judicial Division Division 1 Jury Trial Male Race Non-white Court Significant Variables in Model 3 (p=.05) Extra Conviction Class (E-H) Conviction Type Prior Record Race Charge reduction Judicial Division Percent Poverty Court Jury Trial Significant Variables in Model 3 that DECREASE the likelihood of the judge imposing an active sentence (compared to reference category) Extra Drug or Person Conviction (Compared to Property) Judicial Division (Compared to Division 1) Percent Poverty Significant Variables in Model 3 that INCREASE likelihood of a judge imposing an active sentence (compared to the reference category) Extra 9 Judges do not have discretion in Conviction Classes A-D to impose a non-active sentence. Those classes are not relevant to this model. 17

19 Conviction Class E-H (Compared to Class I) Charge reduction Prior Record Level II-VI (Compared to Level I) Nonwhite Jury Trial Superior Court (Compared to H&I Felony District Court) MODEL 4: Severity of Minimum Sentence Length Imposed Effects of and Extralegal Factors N=13,177 Felony convictions where Active sentence was imposed ( ) Average Sentence Length = 33 months Variables Included in Model 4 Reference Category (where applicable) Conviction Class (B1-H) 10 Class I Conviction Conviction Type Property Conviction Prior Record Level (II-VI) Prior Record Level I Charge Reduction Extralegal Percent Poverty Judicial Division Division 1 Jury Trial Male Race Non-white Court Significant Variables in Model 4 (p=.05) Extra Conviction Class (B1-H) Conviction Type Prior Record II-VI Judicial Division Charge reduction Jury Trial Significant Variables in Model 4 that DECREASE the severity of the minimum sentence imposed Extra N/A Judicial Division 2-6 (Compared to Division 1) Significant Variables in Model 3 that INCREASE the severity of the minimum sentence imposed Extra Conviction Class B1-H (Compared to Class I) Jury Trial Person Conviction (Compared to Property) Prior Record Level II-VI (Compared to Level I) 10 Class A Convictions require mandatory life sentences, therefore they are not relevant to this model. 18

20 APPENDIX E Charged Offense Class Compared to Convicted Offense Class Felony Charges Charged Offense Class Convicted Offense Class B1 B2 C D E F G H I Misdemeanor TOTAL % N B * 536 B C ,364 D ,696 E ,639 F ,753 G ,842 H ,523 I ,367 *Numbers may not add to 100% due to rounding 19