Equitable application of death sentences requires careful monitoring of our abilities to rank homicides along various continua that would differentiate the homicides into ascending levels of severity and aggravation. Nonetheless, no state has instituted a program of data gathering and analysis that would allow neutral parties to do this. Even if states agreed that such monitoring should be done, there are barriers that constrain even the best efforts to rank-order homicides on their severity and “deservedness” of death.
In recent years, America’s death penalty debates have uncovered wide areas of agreement. For example, whether friend or foe of executions, virtually all responsible parties agree that a convicted felon should be punished severely and, in the most extreme kind of case, receive a sentence that ensures that the offender never will be released from prison. (In thirty-seven of the thirty-eight states that today authorize the death penalty, a person convicted of capital murder alternatively can be sentenced to life imprisonment without parole.) Most knowledgeable parties agree that jurisdictions must seek to expand and improve various programs and policies that promise to reduce rates of criminal violence. There is an emerging consensus that the criminal justice system needs to do more to help families of homicide victims. Few would disagree with the assertion of Walter Berns, one of the nation’s most articulate supporters of the death penalty, who argues that regardless of how strongly a person may support the death penalty in theory, the propriety of the penalty in practice “depends on our ability to restrict its use to the worst of our criminals and to impose it in a nondiscriminatory fashion.” Walter Berns, Defending the Death Penalty, 26 Crime & Delinquency 503, 511 (1980).
Acknowledging Fallibility and Unfairness
Over the past three decades, researchers have assembled a massive body of evidence that challenges the assertion that modern capital punishment systems have succeeded in assuring that only the “worst of the worst” are sent to America’s death rows. This evidence can be categorized into two general types: evidence showing the “fallibility” of death penalty decisions and evidence challenging the fundamental “fairness” of such decisions.
By fallibility, we refer to evidence showing that as long as states use the death penalty, at least some innocent defendants will be sentenced and (arguably) put to death. While concern about this risk is not new, the apprehension regarding erroneous convictions was rekindled in the years after Furman v. Georgia, 408 U.S. 238 (1972) . First, in 1987, an article by Hugo Adam Bedau and Michael L. Radelet in the Stanford Law Review documented two dozen cases in which persons sentenced to death since Furman later had been released because of doubts about guilt. The number of known erroneous convictions had grown to 124 by mid-2007, according to the Death Penalty Information Center. Second, in the 1990s, improvements in DNA technology provided the means of proving beyond any doubt that innocent people have been convicted of crimes and some sentenced to death: 206 prisoners by mid-2007, including 14 on death row, according to the Innocence Project. Researchers usually include in these tallies only defendants who were legally and factually found not to be involved in the murders for which they were sentenced to death. Innumerable additional cases of wrongful convictions would be counted if researchers included those who had been convicted of homicide but who in fact had killed in self-defense, by accident, or because of criminal insanity, as well as those convicted of first-degree murder who in fact had possessed the intent or premeditation sufficient only to establish lesser forms of criminal homicide.
At the same time that these kinds of fallibility have become more evident, researchers also have documented a fairly consistent pattern of inequities in the administration of the death penalty that challenges the assertion that the sanction is applied in a consistent and equitable manner. It is still difficult or impossible to predict accurately, among a group of those indicted for or convicted of criminal homicide, who will and who will not be sentenced to death. This concern is reminiscent of the problems with arbitrariness that in part led the Supreme Court to condemn the death penalty statutes reviewed in Furman in 1972. The most consistent extralegal pattern indicates that for similar homicides, and controlling for the most important legally relevant factors, those who kill whites are much more likely to be sentenced to death than those who kill nonwhites. A 1990 U.S. General Accounting Office (GAO) report, Death Penalty Sentencing: Research Indicates Pattern of Racial Disparities, for which the GAO reviewed twenty-eight relevant studies, concluded that in 82 percent of the studies “race of victim was found to influence the likelihood of being charged with capital murder or receiving the death penalty.” Since then, studies in Florida, Illinois, Maryland, California, and Colorado, among others, also have found that those who are suspected of killing whites are much more likely to be sentenced to death than those suspected of committing similar homicides against nonwhite victims.
The racial bias that presumably results in these disparities may have deep psychological roots. One recent study illustrates this point by documenting an implicit preference toward white names and images through the “implicit association test” (IAT). The IAT has been used to assess “socially sensitive attitudes” as an alternative to observational and self-reporting measures of racial preference, and to measure racial attitudes via respondents’ positive or negative associations with racially specific concepts and measures. Several experiments using the IAT “have demonstrated a strong and automatic positive evaluation of White Americans and a relatively negative evaluation of African Americans.” Nilanjana Dasgupta et al., Automatic Preference for White Americans: Eliminating the Familiarity Explanation, 36 J. Experimental Soc. Psychol. 316, 316 (2000). Similarly, a more recent study (using Stanford undergraduates) found that in homicide cases with white victims and black defendants, the more the defendant fits a “stereotypically black appearance” (and controlling for a wide array of other variables), the more likely he is to be sentenced to death. Jennifer L. Eberhardt et al., Looking Deathworthy: Perceived Stereotypicality of Black Defendants Predicts Capital-Sentencing Outcomes, 17 Psychol. Sci. 383 (2006). Stereotypical black traits included size and shape of lips and nose, hair texture, and darkness of skin. This notion of an implicit, and largely unconscious, white preference reasonably can be extended to those involved in the discretionary justice process, from police arrest decisions to discretion in judicial sentencing, from petty crimes to homicide.
Another type of inconsistency in the application of the death penalty results from a pattern where prosecutors and juries in some sections of a state are more likely to seek and impose death sentences than their counterparts in other areas of the state. For example, one study of different areas in Illinois found that the odds of a defendant’s receiving a death sentence in highly populous Cook County were, on average, 83.6 percent lower than those for killing a victim in a similar homicide in rural areas of the state. Geographic variations in death sentencing, due largely to differences in prosecutorial decisions to seek death sentences, also have been found in recent studies in Maryland, California, and Nebraska.
A third type of problem that results in inconsistent application of the death sentence is widespread variation in the quality of counsel available for defendants facing capital charges and resources to mount a vigorous defense that are available to those attorneys. Similarly, researchers have found significant variations in police discretion across communities, as well as substantial differences in departmental investigative capabilities among local police departments.
Understanding the Limits of Data
The persistent evidence of biased, arbitrary, and fallible decision making in death penalty decisions should not come as a surprise. After all, no state systematically collects data on homicides that would allow neutral parties to identify the cases that are most aggravated and most “deserving” of the maximum penalty. At a bare minimum, equitable application of death sentences can be ensured only if there is meaningful proportionality review, which in turn must be based on the systematic collection of comprehensive information on all criminal homicides, from their initial investigation (e.g., the amount of resources that were spent examining the crime scene) through final disposition (e.g., resources for clemency pleadings). All links in the continuous chain of criminal justice and judicial decision making need to provide relevant and accurate information about their role in the process so that their performance and actions can be monitored. Data from all links in the criminal justice decision-making chain are necessary because potentially biased, arbitrary, and/or flawed decisions at one point may well affect subsequent decisions, and yet be invisible to subsequent criminal justice decision makers and, equally important, to outside monitors or evaluators of the system. For example, if researchers begin by gathering data only on who is indicted for capital murder and find no biases in who is sentenced to death, they cannot conclude the entire process is free from bias because it remains possible that local police departments devoted more resources to the investigation of the murder of prominent white victims than they allocated in homicides with nonwhite victims. Similarly, if one studies only those defendants whom jurors sentence to death, then any biases in prosecutorial decisions about which cases are “death eligible” will remain in the darkness.
Given the complexity of developing comprehensive criminal justice information systems capable of monitoring death sentence decisions, the task is, at best, a long-term proposition for virtually all death penalty jurisdictions. The task is complicated further by the structure and geographic organization of criminal justice agencies in the United States. The federal structure of American government and the strong tradition of local control mean that our criminal justice system is divided by level of government (federal, state, and local), geography (state, metropolitan, county, and municipal), and function (police, courts, prosecution and defense, corrections, probation, and parole). These organizational disjunctures make for a fragmented and uncoordinated system of criminal justice agencies, which in turn presents significant barriers to developing integrated criminal justice information systems for state and local governments. Such organizational challenges constrain and limit even the best efforts to construct and implement comprehensive information management systems that collect high-quality data on all criminal homicide cases.
A more challenging problem centers not on jurisdictions’ unwillingness but on their inherent and fundamental inability to systematically monitor potential death penalty cases in a way that would make meaningful proportionality review possible. It may be that no combination of human and information systems oversight would be adequate to ensure that death penalty decision making is sufficiently free of potential bias, arbitrariness, and/or mistakes to make certain that only the worst of the worst offenders are sentenced to death. The unique characteristics of homicide and the death sentence, relative to other crimes and penalties, raise the possibility that powerful political and psychological factors affect the criminal justice decision-making process in ways that would introduce bias, arbitrariness, and/or mistakes into death penalty decisions even if we systematically collected comprehensive information on all criminal homicides and performed well-intentioned proportionality reviews.
A particularly powerful factor in shaping, sensationalizing, and distorting crime reporting is the mass media. In the United States, crime reporting constitutes a significant component of local news reporting, an influence that is unlikely to be neutral or benign. Numerous media researchers have documented that crime news reporting, in general, is more likely to cover homicides involving black suspects and/or white and female victims. Perhaps equally important, if also more subtle, is the finding of a number of studies that news coverage of crime increases fear among viewers, and, according to one study, may even increase public support for conservative law enforcement initiatives. This latter finding is consistent with a very broad body of research that has found that when persons become fearful, they also become more punitive in response to crime. This line of research also has found that increased fear concerning one’s safety and mortality also leads to increased prejudice, stereotyping, and bigotry. Tom Pyszynski et al., In the Wake of 9/11: The Psychology of Terror (2002). Finally, although little research has been conducted that focuses on the media’s impact on criminal justice decision makers, a study found that crime reporting affects how prosecutors negotiate pleas. David Pritchard, Homicide and Bargained Justice: The Agenda-Setting Effect of Crime News on Prosecutors, 50 Pub. Opinion Q. 143 (1986). This study is important because it demonstrates that news media crime reporting very likely influences criminal justice system actors in ways similar to its effect on the general public.
Prosecutors, judges, jurors, and even social scientists are not immune to emotional reactions to homicides, with particular fear or outrage at those homicides that strike closest to home. Our abilities to relate to victims and defendants, and to view them as similar to ourselves, may affect our perceptions of which homicides are the worst of the worst. These problems are potentially aggravated further by research findings that indicate that, for many persons, racial bias and stereotyping are implicit or unconscious. As a result, bias in homicide cases from investigation through judicial decisions, may be unintentional in terms of conscious motives but very real in terms of final results. In practice, these emotional reactions further complicate our ability to use neutral and legally relevant factors to rank homicides on a hierarchy of severity.
Effective Monitoring
While it may not be possible to monitor homicide cases with the perfection needed to ensure that only the worst offenders are being sentenced to death, the possibilities for such a system should not be dismissed before governments show much more willingness than they have to date to gather the various types of data that would begin to make effective monitoring a reality. Even if short of perfection, much more can and should be done. In 1986, Jack Greenberg argued that capital punishment is unconstitutional because contemporary systems of death sentencing are unable to systematically and consistently isolate the worst—and only the worst—criminals for condemnation. Jack Greenberg, Against the American System of Capital Punishment, 99 Harv. L. Rev. 1670 (1986).
Even if we extended the very best efforts of the criminal justice system, it is unlikely that we could design and implement systems that are consistently fair and infallible enough to meet the constitutional standards of a penalty with the finality of capital punishment. This is not to say that death penalty decisions are always biased and/or erroneous, nor is it to say that efforts to enhance criminal justice decision making are not extremely valuable to society. But, given the limits of human monitoring systems and human decision making, erroneous information and/or unconscious bias inevitably will influence at least some death penalty decisions. Such limits must be recognized in considerations of the constitutionality of the death penalty. The implications of bias or mistakes in the application of the death penalty to the overall credibility of the criminal justice system also must be considered. Given the finality of this punishment, even infrequent mistakes in the application of the death penalty will receive widespread coverage and call into question the overall fairness of the system, not to mention the harm and suffering incurred by wrongfully convicted defendants, their families, and the families of the victim. This is the type of long-term cost that cannot be measured in dollars.
Glenn L. Pierce is principal research scientist at the College of Criminal Justice, Northeastern University, in Boston, Massachusetts. Michael L. Radelet is professor and chair of the Department of Sociology at the University of Colorado, in Boulder, Colorado.