Life terms or death sentences: the uneasy relationship between judicial elections and capital punishment.

• Author: Brooks, Richard R.W.

1. INTRODUCTION

   One day Louise Harris approached her lover, Lorenzo "Bo Bo" McCarter, with a proposition that led to an agreement to kill her husband. (1) Harris and McCarter paid Michael Sockwell and Alex Hood one hundred dollars to carry out the gruesome killing. (2) Following the killing, Harris, McCarter, Sockwell, and Hood were all convicted of capital murder in separate proceedings. In each case, a jury recommended life in prison without parole. Yet, in two of the four cases, the trial judge declined to follow the jury recommendations, choosing instead to sentence the defendants to death by electrocution. Different sentences following guilty verdicts on the same offense often occur because the "guilty/not-guilty" determination affords only the crudest approximation of culpability. Through sentencing, however, mitigating and aggravating considerations can give countenance to culpability. Still, the juries in the Harris murder cases had access to these considerations when they reached the same sentence recommendation for all four defendants. Why two defendants ultimately received death sentences and two received life sentences is a question that has more to do with the trial judge's temperament and discretion to impose his preferences than with some "objective" balancing of mitigating and aggravating circumstances. (3) In this Article, we attempt to measure the extent to which judicial temperament affects the likelihood of a defendant being found guilty of murder and the impact of specific judges on sentencing. Using detailed historical data on all murders recorded by the Chicago police over a sixty-year period during the late-nineteenth and early-twentieth centuries, we find significant judge-specific effects for both convictions and sentencing outcomes. Our analysis also reveals a consistent pattern of harsher outcomes correlated with the race of the defendant, (4) the race of the victim, (5) and the killing of police officers. Further, we observe a strong relationship between election years for judges and the likelihood that a defendant will receive a death sentence. That is, conditional on being found guilty of murder, criminal defendants were approximately 15% more likely to be sentenced to death when the sentence was issued during the judge's election year.

   1. POLITICS AND CAPITAL PUNISHMENT

      In Louise Harris's death sentence appeal before the U.S. Supreme Court, Justice John Paul Stevens remarked that judges may be too responsive "to political pressures when pronouncing sentence in highly publicized capital cases." (6) Given the broad support of capital punishment among Americans, Justice Stevens observed that "judges who covet higher office--or who merely wish to remain judges--must constantly profess their fealty to the death penalty." (7) When up for re-election, most judges simply cannot afford to ignore popular sentiment about the death penalty. Nor, apparently, can many other elected officials. (8) For instance, one recent study found that during gubernatorial election years, states are 25% more likely to execute prisoners. (9) Prosecutorial election cycles have also been long known to correlate with the state's willingness to seek death sentences in murder cases. (10) It would be surprising if judges, during their election cycles, were unresponsive to the political pressures confronting their elected counterparts in the governors' and prosecutors' offices. (11) In these offices, the pressure is often relieved through an exercise of gubernatorial or prosecutorial discretion. One might expect that in judicial chambers and courtrooms throughout the country, judicial discretion serves a similar purpose.

   2. JUDICIAL DISCRETION AND JUDGE SPECIFIC EFFECTS

      Judicial discretion can, of course, be good or bad. Through discretion, judges can tailor punishments to conform to socially desirable objectives. Unfortunately, judicial discretion based on unwarranted considerations, such as re-election prospects or ethnicity, will lead to undesirable variations in sentencing. (12) In response to growing perceptions of widespread variations of this sort, the U.S. Congress passed the Sentencing Reform Act in 1984. (13) This Act came nearly one century after studies first identified systemic sentencing variability due to individual differences among judges. (14) One early analysis of discretion in judicial decision-making was offered by George Everson in this journal in 1919. (15) Everson analyzed the sentencing patterns of approximately forty judges on the New York Magistrates' Court: "With the consent of the Chief City Magistrate the Committee on Criminal Courts went over the record of 155,000 or so cases disposed of in 1914...." (16) Significant variations were found: "The results showing to what extent justice is affected by the personality of the judges were so startling and disconcerting that it seemed advisable to discontinue the comparative tables of the records of the justices." (17) Several years later, Frederick Gaudet and his colleagues examined 7442 randomly assigned cases in one New Jersey jurisdiction and observed stark differences among the sentencing behavior of judges. (18)

      To be confident that these observed differences in sentencing were due to the judicial temperament and not other "non-judge" factors that are correlated with particular judges, Everson and Gaudet, et al., relied on random case assignments. Randomly assigned cases will control for these non-judge factors, assuming that the number of cases is sufficiently large. Reliance on randomly assigned cases continues to be a useful methodological device for research on judicial decision-making. For example, Joel Waldfogel recently analyzed data from federal criminal cases that were randomly assigned to ten judges in the Northern District of California from 1984 to 1987. (19) He too found statistically significant inter-judge disparity. Similarly, James Anderson and his colleagues relied on random assignments over 77,201 federal cases from 1981 to 1993 to show a reduction in interjudge sentencing disparity after the implementation of the Federal Sentencing Guidelines. (20) With a sufficient number of cases, an accurate and straightforward comparison of judicial leniency or severity may be made relying on random case assignments. However, the appropriateness of such comparisons is predicated on the validity of the random assignment mechanism. That is, the case assignments must be truly random in order to draw meaningful conclusions. (21) There are many reasons to be skeptical that such mechanisms exist in actual courtroom practice, (22) especially in Chicago during the time of this present study. (23)

      As an alternative to relying on random assignments for valid comparisons, researchers may explicitly control for factors that are correlated with judicial sentencing. One method of controlling for such factors is to place the set of cases being analyzed into subgroups with similar relevant characteristics. Comparisons within subgroups across different judges may then be used to identify judge-specific effects. Edward Green, for instance, employed this procedure in his examination of the Philadelphia criminal court, where he analyzed 1437 cases between 1956 and 1957 among 18 judges. (24) Though Green underestimated his own results, significant disparity among judges was again found. (25) Caution must be exercised when interpreting these results, however, since cases within a subgroup may differ in salient ways that are unobservable to the researchers. Our research is based on simple analysis of variance and multivariate regression techniques, which--for practical purposes--are the same as the subgroup methodology described above, though somewhat more sophisticated. We control for relevant non-judge variables by including them, along with the judge-specific variables, in our regression equations. Unfortunately, just as with the subgroup comparison method, omission of relevant unobserved factors will lead to biased results. Since it is impossible to be certain that all relevant factors are included in the analysis, bias will result if important omitted factors are more likely to occur with a subset of judges.

      Bias resulting from omitted variables will be unavoidable in analyses of judicial decision-making unless the researcher is able perfectly to control the variance among the cases and background factors that judges face. Simulated or mock trials, which evaluate the rulings of multiple judges over the same cases, have been used to evaluate differences across judges in a perfectly controlled setting. The problem with this setting is that the simulation itself may omit some pertinent real-world consideration. To overcome the artificiality of simulations, Shah Diamond and Hans Zeisel looked at New York and Chicago sentencing councils, which consist of a panel of judges who read pre-sentencing reports and offer sentencing recommendations to the presiding trial judge. (26) Judges on sentencing councils avoid the contrived nature of experiments because they are aware their sentencing recommendation will affect real defendants. Using this thoughtful approach, Diamond and Zeisel were able to identify significant disparity among the panel judges in terms of the types of sentences (e.g., imprisonment or probation) and the length of sentences.

      One drawback, however, of using disparity among sentencing councils as a proxy for courtroom variability is that the members of the council are ultimately not issuing the sentence, but rather only making a recommendation from a fairly anonymous vantage point. This anonymity may have implications for sentencing decisions. For example, since sentencing decisions take place in a broader political context than the courtroom, the anonymity of the council may serve to embolden or restrain judicial behavior. On the other hand, trial judges know their decisions often have political resonance, which can affect...