The effect of plea bargaining vs. trial conviction on the sentencing of offenders charged with a drug offense in Cook County, Illinois
v TABLE OF CONTENTS
LIST OF TABLES vii
LIST OF FIGURES ix
CHAPTER 1: INTRODUCTION 1
CHAPTER 2: LITERATURE REVIEW 3
CHAPTER 3: METHODOLOGY 24 Data Description 24 Data Permissions 25 Variables in Dataset 25 Dependent Variables 25 Independent Variables 26 Analyses Performed 29 Coding 31 Schemes for Codes 32
CHAPTER 4: RESULTS 36 Dominant Statutes 36 Probation vs. Prison 38 Conviction Mode 39 Offense Seriousness 41 Gender 43 Gender and Conviction Mode Compared to Sentence 44 Ethnicity 48 Ethnicity and Conviction Mode Compared to Sentence 49 Ethnicity and Offense Seriousness 50 Summary 57 General Linear Models 59 Research Question 59 Grand Summary 65
vi CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS 75 Introduction 75 Data and Analyses 76 Discussion of Findings Related to Trial Tax Existence 77 Discussion of Findings Related to Contextual Variables 78 Discussion of Findings Related to Measuring Associations 79 Limitations 79 Implications and Recommendations for Future Research 81 Conclusion 81
APPENDIX A 82
REFERENCE LIST 88
vii LIST OF TABLES
1. Independent Variables 27 2. Statutes and Their Descriptions 28 3. Example of a Contingency Table 30 4. Summary of Independent Variable Coding Schemes 33 5. Frequency Distribution of Statutes Violated 37 6. Comparison of Probation vs. Prison 38 7. Comparison of Plea Bargain vs. Trial Conviction 39 8. Prison Time by Conviction Mode 39 9. Conviction Mode Compared to Probation or Prison 40 10. Comparison of Sentence and Conviction Mode for Mean Sentence (in Months) 40 11. Convictions for Quantity Compared to Offense Seriousness 41 12. Mean Sentence (in Months) for Seriousness Factors 42 13. Analysis of Variance for Sentence Length 43 14. Comparison of Gender and Conviction Mode 44 15. Comparison of Conviction Mode and Mean Sentence (in Months) 45 16. Analysis of Variance for Gender and Conviction Mode Regarding Sentence 46
viii 17. Comparison of Ethnicity and Conviction Mode 48 18. Comparison of Ethnicity and Conviction Mode for Mean Sentence (in Months) 49 19. Analysis of Variance for Ethnicity and Conviction Mode Regarding Sentence 50 20. Comparison of Ethnicity and Offense Seriousness 51 21. Comparison of Ethnicity and Quantity 51 22. Comparison of Ethnicity and Possession vs. Sales Seriousness for Mean Sentence (in Months) 52 23. Comparison of Ethnicity and Quantity for Mean Sentence (in Months) 53 24. Analysis of Variance for Ethnicity and Quantity and Seriousness Regarding Sentence 54
25. Independent Variables 60 26. Summary of Independent Variable Coding Schemes 62 27. Summary of Linear Model Analysis to Predict Sentence Using All Independent Variables 63
28. Predicting Sentence Summary 67 29. Predicting Sentence by Gender for Possession Crimes Considering Quantity 68 30. Predicting Sentence by Gender for Sales Crimes Considering Quantity 70 31. Predicting Sentence by Ethnicity for Possession Crimes Considering Quantity 72 32. Predicting Sentence by Ethnicity for Sales Crimes Considering Quantity 73
ix LIST OF FIGURES
1. Comparison of sentence means considering conviction mode and gender 47 2. Comparison of sentence means considering ethnicity and possess vs. sell 55 3. Comparison of sentence means considering ethnicity and quantity 56
Traditional wisdom suggests those who lose at trial for a criminal charge receive a heftier prison sentence than those who plea bargain. Plea bargaining reduces strain on the courts, expedites adjudication and may indicate the defendant’s propensity for rehabilitation as they accept responsibility for their actions. Some ask why two people charged with the same crime should receive different sentences based on the adjudication method. The Constitution guarantees the right to a jury trial. Innocent defendants may decide to plead guilty for a sure short sentence rather than risk a trial conviction’s lengthier one. This study using statistical procedures examined 12,786 adjudicated drug crime cases between 2004 and 2007 from the Cook County Circuit Court in Illinois. It sought to determine if adjudication method, plea bargain vs. trial conviction, predicted prison sentence while controlling for independent variables such as ethnicity, gender, statute violated, offense seriousness, quantity and interaction effects.
CHAPTER 1 INTRODUCTION For years, criminal justice experts have debated the existence of a trial tax with regards to sentencing. Many believe that in the case of two defendants charged with the same offense, all other conditions being equal, if one pleads guilty and the other goes to trial and is found guilty, the latter will receive a stiffer sentence than the former. Ulmer and Bradley (2006) define trial tax as occurring when, “. . . defendants are substantially penalized if they exercise their right to a jury trial and then lose.” Bogira (2005) defines trial tax as, “. . . the extra punishment a defendant may face merely by virtue of exercising his right to trial.” The term trial tax has even found its way into official court publications. Illinois appellate courts have acknowledged its existence. The Alabama Sentencing Commission has debated its constitutionality. Bogira (2005) claims the use of trial tax for means of judicial efficiency declaring, “A guilty plea can be wrapped up in approximately 20 minutes, where a jury trial usually takes anywhere from two days to a week.” Does the trial tax actually exist? Controlling for other factors, are defendants who plead guilty more likely to receive leniency than those who go to trial? Among offenders who plead guilty or go to trial for a drug offense, does a disparity exist in their sentences? Assuming we find a disparity, does it remain if we control for other variables such as specific crimes, offense severity, ethnicity or gender?
This study will examine adjudication data from the Circuit Court of Cook County. The arrest, charge and disposition records of almost 13,000 defendants adjudicated for drug related offenses will be analyzed. Statistical analyses will be run to check for differences between groups based on the above noted variables. The results of this dissertation may be used for policy development related to the alleged penalization of the constitutional right to a jury trial. It may also serve as the foundation for a future, more encompassing study of the same topic – perhaps Illinois wide or including other states. This project will be performed in conjunction with and receive assistance from the Institute for Metropolitan Affairs, a public policy institute operating within Roosevelt University.
CHAPTER 2 LITERATURE REVIEW The primary issue for this research relates to whether a trial tax exists, or not. Anecdotal evidence supports the notion that, everything else being equal, those who plead guilty for crime x will receive a lighter sentence than those who got to trial and are then convicted of crime x, hence the term trial tax. The defendant is enticed to accept the pact – plead guilty and receive a great bargain, hence, plea bargain. However, if the criminally charged all have a constitutional right to trial why should they be penalized for exercising that right versus accepting a plea bargain? This dissertation will examine the difference, if any, between convicts who went to trial to those who plea bargained. The importance of this dissertation lies with the question of disparity in sentencing between trial and plea bargain convicts. If no inequality is found, the concerns over the use of a trial tax disappear. If those who plea bargain receive a statistically significant lesser sentence, however, then innocents may be far more likely to plead guilty to avoid a lengthy prison term. Plea bargaining becomes a way of trading the risk of a ten year sentence for the certainty of three to five, regardless of guilt. We will see the risk- reduction theme run throughout this literature review. The oldest recorded use of a form of plea bargaining involves Galileo, the 15th century Italian astronomer. In 1633, he avoided death via the Inquisition by pleading
guilty to supporting Copernicus’ heliocentric theory and agreeing to publicly deny said theory. The judges gave him house arrest and he agreed to recite weekly, penitent psalms, which was a better deal than being burned alive. While far removed from modern American courts, we see the seeds of the notion that pleading guilty upfront results in a lighter punishment. The U.S. Supreme Court declared plea bargaining to be a legitimate legal tool via two landmark cases, Brady vs. United States and Santobello vs. United States. In Brady, the Court noted that a plea helped hold down costs and expedited the judicial docket. In the same case it declared that defendants were entitled to “limiting the probable penalty” by pleading guilty. In Santobello, the Court asserted this acceptance of responsibility indicated a better chance for rehabilitation and should therefore be encouraged. However, there is no research cited in the case supporting this conclusion. Neither case, however, discussed the constitutional propriety of how large the disparity in sentences may be, using plea bargains when the prosecution’s case was weak or “bargaining down” to a lesser charge. In a sweeping 1969 decision, Boykin vs. Alabama, the Court declared that Constitutional right to a jury trial may be waived only if the defendant voluntarily accepted said plea bargain. Boykin, represented by a public defender, never formally, with his own words, accepted a plea bargain for the five robbery charges against him. He and his attorney remained silent while the prosecution presented the plea. The trial judge simply entered the plea without asking Boykin if he understood and accepted it. The Court reversed the conviction even though Boykin’s attorney did not object to the plea. Bargaining “down” came into being as a result of sentence guidelines and mandatory minimums. According to Kinsley (2002) with sentencing discretion limited or
removed entirely by the mandatory minimums, prosecutors found it necessary to allow defendants to plea to a lesser charge in order to get the lesser sentence. He notes that many have argued that when a guilty person plea bargains to a lesser charge, it permits the defendant to escape the legal consequences of their crimes. If they go to trial and are convicted, they will be sentenced according the actual crime committed, not a less serious offence. The flip-side of this action is those charged, but innocent, may plead guilty simply to avoid the continued trauma associated with a criminal charge or to avoid or limit a prison term. We must study plea bargains because of their extensive use and potential for abuse. Kinsley (2002) reports that 95% of all criminal cases in the United States are settled by plea bargain. And when, as part of a plea bargain, innocent people confess to a crime they did not commit, that isn't a breakdown of the system. It is the system working exactly as it is supposed to. If you're the suspect, sometimes this means agreeing with the prosecutor that you will confess to jaywalking when you're really guilty of armed robbery.
Fisher (2003) quotes University of Chicago law professor Albert Alschuler as declaring the rate to be around 90%. The King (2005) study, discussed in this review, places the figure at around 98%. While precise figures are not available, legal experts agree that the vast majority of criminal cases are settled by the use of plea bargaining. Per Fisher (2003), the earliest use of plea bargains in the United States is found in the late 18th century, Middlesex County, Massachusetts. He found strikingly similar elements to what we hear today – overburdened and underfunded courts, lack of judicial and law enforcement resources, and those with a sufficient “purse” being able to hire the best lawyers. In addition, these early plea bargains came from criminal charges against
victimless crimes such as selling or drinking alcohol. “The Middlesex County prosecutor devised a system whereby multiple charges for selling liquor without a license would be dropped to one charge, to which defendants would plead nollo contendere [no contest] and be sentenced to a pre-determined fine and court costs.” Critics of the modern war on drugs would no doubt sympathize with their 18th century counterpart critics. Fisher (2003) claims that by 1900 the criticism of what we today call a trial tax had taken root as the severity of a trial sentence ranged from twice to three times as long as those handed down after a plea bargain for the same crime. Does the trial tax exist? If so, does this disparity affect one group, such as ethnicity or gender, more than another? What types of formal research have been conducted to date? Langer (2006) dealt solely with the improper use of plea bargains by prosecutors in terms of violating jurisdictional penal codes covering plea bargain rules and procedures. Langer performed qualitative, document analysis research by studying landmark cases, state and federal, in which Constitutional issues of lack of due process, as it relates to prosecutorial discretion, was the dominant if not sole ground for the case. He categorized the rights violations into four groups: Right to a Hearing and Knowing the Evidence, Proof Beyond a Reasonable Doubt, Right Against Self-incrimination and the Right to an Impartial Adjudicator. Langer (2006) coined terms for two primary categories of defendants being denied due process via a prosecutor’s authority. He refers to the “de facto unilateral adjudication” in which prosecutors, effectively, solely decide guilt or innocence via coercive plea proposals. As prosecutors decide charges, and in effect sentences if found guilty, the sentence differential often leaves the accused with no reasonable choice other
than to accept guilt. In addition, he describes what he labeled “de facto bilateral adjudication” where both prosecution and defense sidestep the courts to adjudicate by mutual consent. Langer argues that the former violates our fundamental due process rights guaranteed by the Constitution. However, as previously noted, the U.S. Supreme Court sanctioned the waiving of Constitutional rights via Boykin vs. Alabama, if the accused voluntarily accepts the plea bargain. By threatening to take cases to trial where no reasonable jury would find guilt or charging defendants (guilty or not) with crimes that do not reflect the incident in question, prosecutors have created what Langer (2006) calls the informal prosecutorial adjudication system. The quintessential example remains sexual assault versus assault. A prosecutor has virtually no case against a defendant charged with rape. He offers to reduce the charge to assault with a sentence of probation if the accused pleads guilty. Granted, prosecutors cannot force a defendant to accept said plea bargain but this does not change the adjudicatory nature of this arrangement. Once again, however, the U.S. Supreme Court has authorized the process by the case Brady vs. United States, actually stating that defendants were entitled to “limit the probable penalty.” Regardless, this unchecked power of prosecutors, Langer asserts, has led to uncountable numbers of coercive plea bargains. Bibas (2004) examined the outcomes of plea bargaining in civil litigation and found results similar to Langer (2006). We see again this notion of hedging a bet with regards to a result. The defendant may truly owe $100,000 but the plaintiff, unwilling to risk a zero dollar judgment, accepts the $50,000. Conversely, the defendant may owe nothing, but, fearing a large judgment, agrees to pay the smaller amount. Similar to
Langer’s concerns over defacto bilateral adjudication, effectively, the attorneys decide the outcome without input from the judge or jury. Bibas (2004) provides a study filled with both actual and hypothetical cases for illustration. Typical examples include a plaintiff claiming $100,000 in damages with a jury only 50% likely to find the defendant negligent. Both parties should therefore settle for $50,000. Similar to Langer (2006), Bibas used document analysis by reviewing high profile civil cases across the United States for his material. He also interviewed attorneys from each side, plaintiff and defendant, asking non-case-specific questions about the process and how the parties ultimately settled. This study proved an interesting contrast to the criminal case related research addressed elsewhere. Bibas (2004) makes little mention of Constitutional rights, loss of liberty or coercion by prosecutors. While he does note similarities between criminal plea bargains and settling tort cases, he spends most of the research considering the similarity between the self-correcting market place and logic behind plea agreements similar to the above mentioned example. While he reviews bargaining for the best financial bargain, the other studies dealt with bargaining over someone’s liberty. Bibas (2004) notes the similarities, such as plea bargaining being hidden from public view, to the criminal court’s use of plea bargains. Be it neighbors feuding over a damaged lawn or a multi-billion dollar tort claim, as the discussion and settlement occurs outside of the courtroom the public will never know what transpired. Unlike a criminal case, however, they will not even know the outcome. In criminal cases, the sentence becomes public record regardless of the adjudication method. In a civil case settled outside of court, only each party knows the outcome. Another difference is that there are
no public defenders in the civil arena. An indigent criminal defendant may receive a defense attorney, paid a straight salary, to represent 10 or 100 clients. Clearly that attorney has personal incentive to plea bargain. That scenario would not exist in a tort claim where counsel is paid via a retainer or even a percentage of recovered funds. Finkelstein (1975) found evidence of another unanticipated cost of plea bargaining, the “implicit rate of non-conviction” – the proportion of defendants pleading guilty who, in all probability, would have not been convicted in a trial. The researcher determined this unobservable variable by creating a conviction probability. If there exists two federal districts with the prosecutors from one always seeking maximum sentences and the other not doing so, logically, defendants in the first district are more likely to plea bargain than those in the second. If every defendant pleading guilty would have been convicted at trial, it stands to reason that, over time, the proportion of acquittals between the two districts would be relatively equal. If, however, the first district sees substantially fewer acquittals than the second, undoubtedly, some of the plea bargaining defendants in the first district would not have been convicted if they had gone to trial. The study tested for a statistically significant correlation between the percentage of plea bargains and acquittal probabilities. The author says little about where he obtained the records or how he coded them. He does note that the data came from the Annual Reports of the Attorney General. Beyond that he notes two limitations to the study. First, he did not control for the variety of criminal cases per district. If one district’s docket saw a disproportionate amount of a particular type of crime, that may skew the results. In addition, presumably those districts more apt to plea bargain, focused their resources on trials and were more likely
to obtain a conviction because they had the time and money to properly litigate. His study dealt only with federal cases – the situation may be different in state criminal courts. Using the Annual Reports of the Attorney General from the twenty-nine federal district courts, Finkelstein (1975) found a statistically significant, strong negative correlation between the percentage of non-convictions (the defendant would probably have been acquitted if they went to trial) to the percentage of guilty pleas. It should be noted that the data in each district was analyzed in the aggregate to provide for a larger sample and greater stability. The data points do not make a perfect line. This could be interpreted as indicating that the non-conviction rate is influenced by variables other than plea bargain rates. However, the correlation equaled -0.849. Using the least squares technique produced a -0.691 slope indicating a non-conviction rate of about 69%. Finkelstein calls this evidence that, “. . . pressures to plead guilty have been used to secure convictions that could not otherwise be obtained.” (Finkelstein, p. 309) Finkelstein (1975) comments on Boykin vs. Alabama, where the U.S. Supreme Court said a defendant may waive his right to a jury trial only if it is done so voluntarily. He notes that while the Court approved this practice, it did so, “. . . only on the assumption that defendants who were convicted on the basis of negotiated pleas of guilt would have been convicted had they elected to stand trial.” (Finkelstein, p. 293) The above data seems to refute that point. Finkelstein goes on to note the temptation defendants, undoubtedly some innocent, must feel to accept a plea – again, we see evidence of bet hedging. The accused accepts a lower sentence in exchange for avoiding a possible, long sentence. He notes, “. . . prosecutors may be using threats of lengthy
sentences and other plea inducing practices to obtain convictions in case in which the government’s evidence is quite insubstantial.” (Finkelstein, p. 293) Moreover, Finkelstein (1975) argues that, what Langer (2006) would call unilateral defacto adjudication, that is, the prosecutor coercing a plea bargain, negates Boykin vs. Alabama in that, “. . . when strong pressure is necessary to compel a confession in a weak case, the prosecutor’s zeal to obtain a conviction by “consent” begins to collide with the defendant’s privilege against self-incrimination.” (Finkelstein, p. 294) Finkelstein also agrees with Langer’s assessment in that, “It appears that informal, and less visible, administrative practices have been used to induce convictions by “consent” in a significant number of cases in which the protections of the formal system would have precluded a condemnation.” (Finkelstein, p. 311) The Finkelstein data analysis for the 1908 to 1928 period found a statistically significant correlation of -0.812. The correlation value is similar to the 1970 – 1974 dataset, but, the slope of the least squares line equaled -0.265 meaning that the non- conviction rate equaled about 26.5%. This slope is only 38% of the value of the first dataset indicating that while plea bargains resulted in the conviction of those who may otherwise have been acquitted, the impact was not as great for that time frame. The 1954 – 1974 dataset revealed a striking correlation of -0.977 with a slope of least squares equal to -0.791 which is higher than the percentage for 1970 – 1974. It should be noted that Finkelstein (1975) does agree with a comment from Brady vs. United States with regards to plea bargains expediting the court docket and saving funds. He wrote, “If insistence on a trial cost nothing, presumably few of the accused
would plead guilty and forgo the chance for a dismissal or acquittal.” (Finkelstein, p. 293) The Finkelstein (1975) study offers compelling evidence that many who plead guilty may actually have been acquitted at trial. Of course, acquittal does not mean they are innocent. Perhaps the guilty received a deserved conviction; perhaps some innocent people bent to the pressure of a prosecutor. The next step would be to determine if a disparity truly exists for sentencing of plea bargained defendants and those who go to trial. Ulmer and Bradley (2006) focus on violent crimes and the practice of plea bargaining. They did this because violent crime charges are more than twice as likely to go to trial, roughly 7%, versus 3% overall, in Pennsylvania. As previously noted by Fisher (2003), King (2005) and Kinsley (2002), the vast majority of charges are settled by plea bargains. Using data from 1997 to 2000 from the Pennsylvania Commission on Sentencing (PCS), their study tested eight hypotheses, two of which mirrored the research in this dissertation. One hypothesis declared, “Among convicted defendants, those convicted by jury trial will be sentenced more severely than those convicted by guilty plea” (Ulmer & Bradley, 2006, p. 637). Another sought to test whether or not, “The jury trial penalty . . . will be significantly greater among those with more extensive prior criminal records” (Ulmer & Bradley, p. 639). The other hypotheses dealt with influence of course caseloads on plea bargaining and issues related to violent crimes, neither of which will play a part in this research.
Ulmer & Bradley (2006) used sentence length as the dependent variable, coded as probation or incarceration with the number of months and offense severity, prior record, plea bargain and trial (bench or jury) as the independent variables. Their large dataset of n = 8,585 allowed for the employment of Hierarchical Linear Modeling (HLM) to create predictor models and search for group variations using separate but interrelated units of analysis. Level 1, individual case / defendant, characteristics were nested and could interact with Level 2, county courts, individual and cross-county (statewide) outcomes, that is, the sentence length. The authors admitted several limitations to their study, such as not measuring variables such as the socio-economic status of the subjects, whether they had public or private defense attorneys, victim characteristics (assuming harsher sentences if a child, female or elderly person fell target) or pretrial release status. Ulmer and Bradley (2006) noted that individuals held in jail, awaiting adjudication, often did not serve prison time. That is, if someone cannot afford or does not receive bail and they remain in jail for a significant period of time, if they would have received a sentence of X months, the judge could simply release them with time served. While that time would count towards the sentence if the defendant were sentenced to additional time, it does not appear in the Pennsylvania Commission on Sentencing dataset as incarceration time. Obviously, the fact that they studied data from only one state limits the results. The results of this study support the existence of a trial tax. Ulmer and Bradley (2006) found that, “. . . the odds of incarceration following a bench trial are roughly 2.2 times the odds for guilty plea, while a jury trial conviction has roughly 2.7 times the incarceration odds of a guilty plea” (Ulmer & Bradley, p. 650). They concluded that,
“Overall, then, conviction by trial, especially jury trial, carries a meaningful additional sentencing penalty . . .” (Ulmer & Bradley, p. 650). The results of this study do support the hypothesis that criminal history plays a role in the sentence depending on whether or not the defendant pleas or goes to trial; however, not in the correlation one might expect. Ulmer and Bradley (2006) determined that, “. . . the jury trial penalty decreases as the prior record score of the defendant increases” (Ulmer & Bradley, p. 653). The authors ran separate HLM models to compare jury trial sentencing results for subjects with high criminal record scores to those with low criminal record scores. The authors found that the difference between having many prior convictions and going to trial or not was not statistically significant in terms of the odds of incarceration. However, subjects with low criminal record scores had 3.5 times the probability of incarceration if they had chosen a trial vs. plea bargain. Oddly, the jury trial tax seemed heavier for subjects with less criminal histories than those with an extensive record. Agreeing with Finkelstein (1975) and Brady vs. United States, the authors note that, “Most researchers argue that rewarding those who plead guilty and penalizing those who lose at trial reflects the need for efficiency in case processing.” (Ulmer & Bradley, 2006, p. 635) However, unlike Finkelstein, the authors offer support for this scenario in that, “Rewarding those who plead guilty with lighter sentences is widely seen as necessary to encourage defendants’ ‘remorse,’ ‘acceptance of responsibility’ for crimes . . . losing [at trial] may signal a defendant’s lack of remorse, and therefore greater blameworthiness, to judges.” (Ulmer & Bradley, p. 636) Similar to Santobello vs. United
States, accepting responsibility may be evidence of a better chance of rehabilitation, thus, the lighter sentence. Johnson (2003) used the Pennsylvania Commission on Sentencing (PCS) data to focus on departures from sentencing guidelines, not the complete sentence itself, and then looked for differences between groups, depending on the type of conviction. A departure, in criminal sentencing terminology, refers to an allowed increase or decrease in a sentence range established by guidelines or a sentence set by statute. For example, if a guideline calls for a minimum sentence of x months for a crime, the judge may be allowed to assign a downward departure of y months for showing remorse and acceptance of responsibility. That defendant’s sentence would then be x – y months. Or, if the maximum sentence for some crime equaled x months, but the convicted person acted as ring-leader, the judge may be allowed to assign an upward departure of y months. The sentence then becomes x + y months. The author used PCS data from 1996 – 1998 to examine the likelihood of receiving a sentence that departs from the guidelines. He then searched for disparities based on conviction modes and ethnicity. The author defined the four modes as non- negotiated pleas where the accused pleads guilty without discussion, negotiated pleas where the defense and prosecution negotiated a plea agreement, bench trials where the case was tried by the judge and jury trials where a panel of jurors decided to convict or not. The researcher formed six hypotheses. The first sought to test the question of sentence outcome disparity based on ethnicity. The second focused on sentence disparity