Our study results identify a series of conditions associated with high rates of capital error. Every such error is serious, casting substantial doubt on the reliability of a death verdict. Together, all such error—which was found in 68% of all capital verdicts imposed and fully reviewed between 1973 and 1995—creates a high risk that innocent people have been executed, and will continue to be executed unless major policy changes are made.

A central goal of this decade-long study and the detailed description of its findings here and in A Broken System, Part I is to suggest policy options for responding to serious capital error and the resulting risk that innocent people will be executed. We divide our discussion of these options into three sections. Section A discusses the only two options that promise substantial declines in chronic capital error rates and the resulting risk of executing the innocent: severely curbing the scope of the death penalty to reach only the small number of offenses as to which there is a broad social consensus that only the death penalty will serve, or ending the death penalty entirely. The next section discusses options that will likely have less impact on serious capital error and the risk of executing the innocent but, especially in combination with each other, can help moderate the problem. The final section discusses options to avoid because they might magnify serious error.

We present the reforms discussed below as options among which death penalty states and counties might choose based on the particular risk of serious capital error each jurisdiction faces from the 10 or so risk factors our study identifies and from other factors. In three instances, we identify options as close to a policy imperative in the states and counties to which they apply.⁸⁷²

Information on the size or intensity of each of the 10 risk factors in each of the 34 study states during the 1973-1995 period can be found in:

• Table 18, pp. 344-45 above; and

• the accompanying tables in notes 774, 788, 797, 842 above.

  Additional comparative information on the 34 study states that fills out their respective risk profiles in regard to serious capital error is in:

• Figures 1A, 1B, 9, and 11-21, pp. 50-51, 79, 121-32 above; and

• Appendices A (state report cards), C (state post-conviction cases by state) and D (examples of federal habeas cases).

Similar information on the 244 American counties that capitally sentenced five or more people during the study period, and less extensive information on all 1004 counties that imposed at least one capital sentence, is presented in:

• Figures 42A and 42B (map of all capital counties), pp. 248-49 above;

• Tables 10-16, 19 and 20, pp. 290-91, 294-95, 297-99, 301, 304, 346 and 348 above; and

• Appendix B.

These and other existing sources may not provide enough information to comprehensively assess local conditions in particular states and counties. For that reason, it is strongly recommended that states and counties undertake their own comprehensive analyses of local death-sentencing practices, problems and conditions—as Arizona, Illinois, Indiana, Maryland, Nebraska, Nevada, North Carolina, Virginia and the U.S. Department of Justice have recently done.⁸⁷³

A. Attacking the Problem by Severely Curtailing the Death Penalty

As we develop in Part II above, serious capital error is a chronic problem with severe effects and monumental risks, including that innocent people will be executed. What is worse, rigorous examination reveals that some of the conditions that are most strongly associated with capital error are not easily changed.

Underlying several of the important explanatory conditions revealed by our regression analyses are fears about serious crime that generate pressure on officials to extend the death penalty to cases that are not highly aggravated—where the risk of error, reversal, retrial and a non-capital retrial verdict is high.⁸⁷⁴ Among the sources of such fears are well-founded doubts about some states' response to serious crime given their poor records of apprehending, convicting and punishing serious criminals.⁸⁷⁵ Racial conditions—how heavily the threat of homicide is concentrated in the white as opposed to the black community, and the relative size of the black community—are other apparent sources of pressure to use the death penalty in the less discriminate manner that our findings associate with a high risk of error. Even the size and density of a state or county's population are linked to the amount of serious error discovered at various stages of review.⁸⁷⁶ Obviously, none of these conditions—the problem of serious crime,⁸⁷⁷ the racial makeup of communities, the distribution of homicides between whites and blacks, and the distribution of the entire population between urban and rural areas—can easily be changed in an effort to directly alter the conditions associated with high capital error rates.

Moreover, even where changes can have a direct impact on potentially causal conditions that our analyses identify—e.g., amending state constitutions to replace judicial elections with appointment for life,⁸⁷⁸ or increasing funding for overburdened courts⁸⁷⁹—are unlikely to occur any time soon. To make matters worse, our analyses indicate that still other factors are at work that are registered by our general measure of time trend and are associated with large increases in error rates over time.⁸⁸⁰ Because the exact contours of these latter influences are unknown—except for the fact that they are getting worse—it is possible that the ameliorative effect of reforms aimed at known risk factors could be overwhelmed by the intensifying effect over time of less well-understood forces.⁸⁸¹

As the next section indicates, therefore, the best that can be done is to try to narrow the available capital-sentencing outlets for pressures associated with race, crime and politics by more firmly restricting the penalty to highly aggravated cases. The uncertain prospects for this approach are indicated by a separate study finding: The conditions that most strongly predict capital error operate mainly at the level of state and county death-sentencing policy, not at the level at which policy is applied to individual cases.⁸⁸² The actions most associated with capital error are those broadly defining the classes of cases and threshold amounts of evidence of guilt and aggravation that qualify for capital charging and sentencing—inclusive definitions of capital murder, long lists of aggravating factors or excessively encompassing factors, and open-ended interpretations of those definitions and factors on a statewide basis by state supreme courts and state's attorneys and on a county-wide basis by local sentencing courts and district attorneys.

This finding suggests that it will be difficult to design effective capital reforms. Inevitably, the officials whose policies must be relied upon to moderate the risk of dangerous error by limiting opportunities to use the death penalty in marginal cases are the same officials who have been susceptible in the past to pressures to expand the death penalty, with the devastating and chronic effects that we describe above.⁸⁸³

Because harms from chronic capital error are great, the causes cannot be directly rooted out, and indirect remedies are unreliable, some states or counties might conclude that the best answer to chronic capital error is to stop using the death penalty entirely, or to limit its use to the small number of offenses as to which there is something approaching a social consensus that only the death penalty will serve. For some states, the events of September 11, 2001, in New York City, Washington, D.C. and Pennsylvania, and events in Oklahoma City in April 1995, may provide a benchmark for defining those offenses.

B. Moderating the Problem Through Targeted Changes in Capital Policy and Practice

Some states and counties will want to explore less comprehensive reform options before concluding that the penalty's costs are not worth incurring. After discussing the overriding goal of ameliorative policies, this sections lists policy options aimed at entire sets of cases, then options applicable case by case.

1. The central goal of moderating reforms: limiting the death penalty to very highly aggravated cases.

The goal of the collection of reforms discussed below is dictated by our principal finding: The more states and counties use the death penalty, and the more often they extend it to cases that are not highly aggravated, the higher is the risk that any death verdict they impose will be reversed due to serious error. Additional findings supplement this main conclusion by identifying conditions that pressure officials to adopt over-broad death-sentencing policies (crime fears based on inadequate law enforcement policies, relatively high concentrations of homicides in the white community relative to the black community, and large African-American populations) or the mechanisms through which these pressures are communicated to officials (direct election of judges and other officials, controversies surrounding reversals of death verdicts in non-urban areas and complaints about backlogs of capital cases awaiting review). As we discuss above, however, these contributing factors are difficult to eliminate directly,⁸⁸⁴ and probably are only a subset of the conditions prompting excessive capital sentencing.⁸⁸⁵ As a result, reform efforts can best address the problem by attacking the crucial condition that links these other conditions to high error rates and is itself linked to a high risk of error: excessive death sentencing in cases where aggravation is not strong.

Therefore, the central goal of these targeted reform options is to limit the death penalty to "the worst of the worst"—to defendants who can be shown without doubt to have committed a murder characterized by high concentrations of undeniably aggravating circumstances. Accomplishing this goal requires firm policies that (1) remove the death penalty from consideration in cases where the evidence of guilt or the level of aggravation is not strong but where law enforcement or other crises create powerful pressures to apply the death penalty broadly, and that (2) insulate death-sentencing policies and decisions from direct political pressure as much as possible.

Strong support for these conclusions is found in the principal finding of the first comprehensive, state-specific study of the death penalty to be completed in the recent period of concerns about capital punishment. As the chief researcher on that study, University of Iowa Law Professor David Baldus, testified to the Nebraska Legislature, which commissioned the study, "[i]f Nebraska would limit asking for the death penalty, . . . there would be fewer questions about application of the penalty" and less reason to fear that "race or wealth of the criminal or victim was a factor in receiving the death penalty."⁸⁸⁶

Some of the options laid out below focus on the guilt-innocence decision in potentially capital cases. Others focus on the sentencing decision. Both sets of options would likely decrease the risk of serious error in both types of decisions. Our findings indicate that it is the goal of securing death sentences in cases where the evidence or aggravation is not strong that invites the application of unreliable policies and practices. That invitation to error extends to both phases of trial, because both are crucial to obtaining a death sentence. If police, prosecutors, trial judges and jurors follow policies that discourage capital convictions in cases where the evidence is weak, that of course will also discourages capital sentences in the same cases. Likewise, if trial actors follow policies they know make a death sentence unlikely because the evidence, level of aggravation vs. mitigation or law does not clearly warrant that sentence, their temptation to cut corners or use over-broad policies to secure a capital conviction as a prelude to a death sentence also declines.⁸⁸⁷

2. Ten ways to moderate error by modifying capital policy and practice.

This section lists 10 ways to lower the risk of serious capital error, starting with six options for categorically moderating the number of capital convictions and sentences.

a. Require proof beyond any doubt that the defendant committed the capital crime.

In a speech to the National Press Club last year, Oklahoma's Republican Governor Frank Keating, a strong death-penalty supporter, proposed limiting the death penalty to defendants found beyond any doubt to have committed a capital crime. Conservative columnist John Podhoretz made a similar proposal in the New York Post as a way to "save the death penalty."⁸⁸⁸ Our study results provide strong support for these proposals. Our analyses show that:

• The crucial condition putting states and counties at risk of high capital error rates is high rates of capital convictions and sentences per 1000 homicides.

• The crucial condition putting particular capital verdicts at risk of error is their imposition where the case for convicting the defendant of a capital crime, or for imposing the death penalty, is not strong.

• The worst effect of serious error is convicting and potentially executing innocent persons.

• Appellate courts are not a failsafe protection against these harmful outcomes, given their record of affirming the convictions of innocent death row inmates—including where they acknowledged that the evidence was weak but affirmed anyway because the defendant could not then prove he was probably innocent.⁸⁸⁹

Under these circumstances, an especially effective way to discourage death verdicts where the risk of error is great is to eliminate cases where there is doubt about the defendant's guilt. This would replace the rule in most states that jurors who harbor doubts about a capital defendant's guilt must still convict him if they think his guilt is "beyond a reasonable doubt."

In many capital cases, fingerprints, DNA, statements by the victim before dying, uncontradicted and corroborated eye-witness accounts, confessions under non-coercive conditions, and the defendant's possession of information or property that only the killer could have, leave jurors with no doubt about who killed the victim. Faced with this kind of evidence, some capital defendants admit guilt at trial, arguing for a lesser sentence based on extenuating circumstances. In other cases, however, the evidence is contested and unreliable—most classically, when it is based on testimony of jailhouse informants who profit by claiming the defendant confessed to the killing, questionable forensic techniques such as hair and fiber analysis and uncorroborated eye-witness identifications made under stress and describing fleeting events that were difficult to see. Faced with such evidence, but also with an egregious crime and a desire to protect the community, jurors told to decide whether the evidence is "beyond a reasonable doubt but not beyond any doubt" may convict innocent defendants because they are convinced the crime requires severe punishment, even if they are not sure the defendant committed it.⁸⁹⁰

The "beyond a reasonable doubt" standard also invites error because courts have been unable to narrow the kinds of doubts that jurors may harbor and still convict. Many courts have given up trying to define "reasonable doubt" and left it to jurors to decide for themselves whether doubts they harbor are "reasonable."⁸⁹¹ The result in capital cases is that jurors faced with evidence that whoever committed the crime poses an intolerable threat to the community, and who believe the defendant is the only suspect, may conclude that the "reasonable" thing to do is convict the defendant and sentence him to die despite lingering doubts about his guilt.

When lengthy or permanent imprisonment is the result—allowing mistakes to be corrected whenever they are discovered—the risks from using the ambiguous "reasonable doubt" standard are justified. The same risks are less sensible when the penalty is death, and mistakes are not correctable. This is especially so given the connection our analyses reveal between death-sentencing policies encompassing relatively weaker cases and a higher risk of serious, reversible error. Those findings counsel against using the death penalty when jurors and reviewing courts have doubts about the defendant's guilt, because it is in just such "close" cases where the probability of serious error is the greatest.

b. Require that aggravating circumstances substantially outweigh mitigating ones and warrant death before a death sentence may be imposed.

Capital jurisdictions are split over how aggravated a first-degree murder must be, after accounting for mitigating circumstances, before the death penalty may be imposed. A minority of jurisdictions bar the death penalty unless aggravating circumstances substantially outweigh the mitigating circumstances, and unless the jury is convinced that the death penalty is demanded by the high degree of aggravation that remains after mitigation is considered. These jurisdictions thus require jurors to impose a lesser sentence unless they are convinced that the case is so aggravated, after taking mitigating factors into account, that only the death penalty will suffice to punish the offender and protect society. By limiting the death penalty to the strongest cases for that punishment, these policies are well calculated to avoid the high rates of unreliable error that our regression analyses associate with broad death-sentencing policies.

In other states, such as California and Pennsylvania, jurors are told that they must impose death if the aggravating circumstances outweigh the mitigating circumstances by any amount, however minor or minuscule.⁸⁹² And still other states, such as Arizona, require the death penalty when aggravating and mitigating factors are evenly balanced—i.e., unless mitigating factors outweigh aggravating ones by some or even a significant amount.⁸⁹³ Kansas also followed this practice until late last year when its supreme court concluded that "fundamental fairness" demands that ties should go to the defendant when life is at stake.⁸⁹⁴ It is hard to imagine sentencing policies that are more likely than those of California, Pennsylvania and Arizona to inflate capital-sentencing rates—and, as a result, rates of serious capital error—through imposition of death verdicts in marginal cases. A simple way to moderate this risk is to limit death verdicts to clear cases—those where the jury finds that aggravation so far exceeds mitigation that only a death verdict can suffice—and to adopt model jury instructions that clearly inform jurors about the findings needed to permit a death verdict.⁸⁹⁵

c. Bar the death penalty for defendants with inherently extenuating conditions.

States also may moderate death-sentencing rates and the resulting risk of serious error and of convicting and condemning the innocent by barring capital prosecutions of defendants with inherently mitigating conditions, especially conditions that keep defendants from effectively defending themselves against false charges or from showing that the evidence and law do not permit their execution.

i. Mentally retarded persons. Because mentally retarded defendants are inherently weak candidates for the death penalty, their cases are prime candidates for serious capital error, reversal and retrial, and they are especially at risk of being convicted and condemned despite being innocent. This explains the strong trend recently towards banning execution of the mentally retarded.⁸⁹⁶ Even strong death penalty supporters such as Criminal Justice Legal Foundation legal director Kent Scheidegger acknowledge the appropriateness ">[a]s a matter of policy . . . [of] a general rule that says we're not going to execute someone who is in fact retarded.'"⁸⁹⁷

Mental retardation is inherently mitigating because the condition decreases criminal blame for reasons beyond the control of the defendant and makes it more likely that the defendant was led into criminal activity by others more responsible than he or she. Moreover, as several recent exonerations establish—including those of Earl Washington and Anthony Porter, the defendants in two of our four case studies of capitally convicted and sentenced individuals whose death verdicts were upheld at all three levels of appellate review though they in fact were innocent—mentally retarded individuals are less capable than other defendants of helping their lawyers prove their innocence.⁸⁹⁸ In some cases such as Washington's, retarded individuals are actually led into confessing to crimes they could not have committed.⁸⁹⁹

ii. Juveniles. The Constitution bars executions for offenses committed by children 15 years old or younger,⁹⁰⁰ and over a dozen states and the federal government ban the death penalty for offenses committed by teenagers below the age of 18. Several other states are considering such bans.⁹⁰¹ Barring the death penalty for crimes committed by juveniles is another logical way to lower capital error rates by removing inherently marginal cases from capital eligibility. Youth is a strong and well-recognized basis for mitigation for many of the same reasons as retardation and also because of the greater chance that defendants who committed serious crimes before reaching maturity can be reformed by long prison terms, until they are no longer a danger to the community.⁹⁰²

iii. Severely mentally disordered defendants. Severe mental disorder is another long-recognized basis for mitigation, and another condition that prevents defendants from helping to prove their innocence or that they are unfit candidates for execution.⁹⁰³ Capitally trying such defendants is extremely expensive, given the many points during the trial when medical and psychiatric examinations, neurological tests and battles of experts are required in order to answer a lengthy series of vexing questions that determine the appropriate legal disposition of such cases: Was the defendant competent to waive his rights and confess or submit to other investigative procedures by police officers? Was he competent to stand trial—does he understand the proceedings; can he assist his lawyer? Was he insane or was his capacity "diminished" at the time of the crime? Did his mental disorder substantially impair his capacity to understand his actions and conform them to the law? Is he competent to be executed?

Death verdicts imposed on severely mentally disordered defendants are also unusually expensive to review, given the need for additional costly examinations and testimony by qualified physicians and mental health professionals. And such cases present many more reasons why death verdicts may be overturned, given the wide range of legal challenges that arise solely because the defendant is severally mentally disordered—e.g., that the defendant was coerced into confessing to mental health professionals who examined him for the state, was incompetent to stand trial, was not given necessary expert assistance and tests, was improperly found to be sane and to lack mental disorders that are a defense to the crime charged or provide a basis for a lesser sentence, is mentally unfit to be executed, and most importantly was incompetently represented by a lawyer with no experience handling the special problems presented by a severely mentally disordered capital client.⁹⁰⁴

Many of these costs and bases for reversal arise in cases in which there is agreement among mental health professionals employed by the state as well as the defendant that the accused suffers from a well-recognized and medically accepted psychotic mental disorder, such as paranoid schizophrenia, with symptoms that strongly mitigate (e.g., reduced culpability, and a susceptibility to treatment and rehabilitation). The reason high costs begin, rather than end, at this point is that, under current law, even uncontradicted proof of this mental condition is not a defense to a capital conviction or sentence. Instead, it triggers numerous intricate bodies of legal rules stating when, despite the defendant's severe and well-recognized condition, prosecutors may nonetheless bring capital charges against him; when trial judges may nonetheless deem the defendant competent to have submitted to certain investigatory procedures and to be brought to trial; when jurors may nonetheless convict him and sentence him to die; and when the state, thereafter, may conclude that he is sufficiently competent to be executed.

A consensus on the existence of a psychosis by panels of mental health professionals thus simultaneously (1) identifies the small minority of potentially capital cases in which there is no doubt that a true and inherently mitigating mental disorder is present; (2) distinguishes the large majority of cases in which any mental problems that may exist do not rise to the level of a clearly proven psychosis; and (3) identifies the case as one bound to be extremely expensive—requiring the employment and compensation both of medical or psychiatric specialists and a specially trained capital defense lawyer—and as a prime candidate for lengthy review, probable reversal and a costly retrial. States or counties that bar capital prosecutions when there is clear proof of psychosis or other severe mental disorders thus stand to avoid many of the worst capital costs and risks of serious, reversible error.

d. Make life imprisonment without parole an alternative to the death penalty and clearly inform sentencing juries of that option.

Recently, respected criminologist and death penalty advocate James Q. Wilson argued that capital juries be given life-without-parole options to the death penalty, so that "jurors who may have some doubts about the strength of the evidence or some other plausible worry [may] hedge their bets [by sentencing the defendant to life without parole] if they are so inclined."⁹⁰⁵ Former federal judge and FBI Director William Sessions also recently proposed that jurors be "able to impose sentences, short of death, that they believe will protect society from the criminal."⁹⁰⁶ Providing support for these views, analyses show that (1) jurors are capable of identifying offenders for whom the death penalty is not warranted as long as there are strong assurances that the offender will remain in prison until he is no longer a threat to society, but that (2) jurors usually will not impose life verdicts in such cases—even though they believe the death penalty is not required—unless they are assured by the trial judge that the defendant will not be eligible for parole.⁹⁰⁷

These findings identify two steps that together can effectively discourage death verdicts in cases where our analyses show the risk of capital error is especially high because the case is not "the worst of the worst": (1) adopt life without possibility of parole as an alternative to the death penalty, and (2) require judges to clearly inform jurors of that option. Our analyses predict that these steps will be associated with lower capital error rates for two reasons: They promote lower capital-sentencing rates by excluding marginal cases where jurors believe a lesser sentence will suffice, and they increase incarceration rates for murder which is itself a condition associated with lower capital error rates.⁹⁰⁸

e. Abolish judge overrides of jury life verdicts.

If there is a single capital policy that most enhances the risk of serious capital error identified by our regression analyses, it is the authority four states—Alabama, Delaware, Florida and Indiana—give trial judges, after jurors vote to impose a life sentence, to override that decision and impose death.⁹⁰⁹ With the exception of Indiana, these states, not surprisingly, are in the top rank in terms of the number of death verdicts per 1000 homicides they impose.⁹¹⁰ Overrides are especially risky in Florida and Indiana because the judges who decide whether to override jury verdicts are subject to popular election. They are most especially risky in Alabama, which not only places particularly heavy political pressure on its elected judges, but also gives them total discretion to override jury verdicts for any reason, without explanation.⁹¹¹

Jury verdicts in capital cases allow 12 citizens to express the community's judgment on whether the death penalty is appropriate for the offense.⁹¹² When jurors hear all the evidence of aggravating and mitigating factors and vote to impose a life sentence, they identify the case as, by definition, a weak candidate for the death penalty. A jury override policy thus gives prosecutors two chances to convince a sentencer to impose a death sentence in inherently weak cases. The policy also puts political pressure on elected judges to substitute their judgment for that of jurors who represent community values at least as well as the judge but are not politically vulnerable. Because jury overrides are an explicit policy of imposing additional death sentences in what by definition are weak cases, and because they are susceptible to political pressures on judges whose re-election prospects are tied to their override records,⁹¹³ the resulting death verdicts fall simultaneously in several categories in which the risk of serious capital error is the greatest.

This conclusion is supported by findings of other researchers that:

• When placed in the hands of politically vulnerable judges, a mechanism that originally was expected to afford capital defendants "a second chance for life with the trial judge,"⁹¹⁴ is far more frequently used to impose death verdicts in cases where the conscience of the community has found the evidence too weak to justify that penalty;⁹¹⁵

• The more political pressure a state's judicial selection method and other conditions place on judges, the more likely it is that judges will exercise their override power to impose death sentences in cases where juries believed death was not warranted.⁹¹⁶

• In some states, there is evidence that death verdicts imposed by judges in cases where the jury voted for life are especially likely to be overturned on appeal.⁹¹⁷

Whatever one might think about reforms keeping jurors from imposing death sentences they otherwise might have imposed, there is almost nothing to be said for a policy that routinely generates death sentences in cases in which a majority of jurors have voted against the penalty. More so than is true of any other reform, our study findings support an end to judge overrides of jury verdicts for life.

f. Use comparative review of murder sentences to identify what counts as "the worst of the worst" in the state, and to overturn outlying death verdicts.

Our findings indicate that over-broad capital charging, convicting and sentencing policies force capital appellate judges to function as substitute capital sentencers to winnow the many capital verdicts imposed at trial down to the few death verdicts the evidence and aggravating circumstances clearly warrant.⁹¹⁸ A problem with relying on appellate judges to perform this task is that they for the most part are limited to reversing cases in which there not only is an improper bottom-line outcome—a death verdict the evidence, circumstances or law do not permit—but also some procedural error that is banned by state or federal law. Absent a procedural error, appellate courts usually are not authorized to cure even very serious mistakes in the bottom-line outcome of the trial.⁹¹⁹

Most state statutes, however, invite state direct appeal judges to conduct one type of review of the bottom-line validity of the capital outcome—by comparing cases in which death verdicts are imposed in the state to others in which they are not imposed to make sure that the death penalty is being used consistently across the state. Regrettably, almost no state appellate courts attempt to assure sentencing consistency in like cases by comparing murder cases in which the death penalty is imposed to those in which it is not imposed.⁹²⁰ Our analyses suggest that by neglecting comparative review, state high courts surrender an important opportunity to identify what prosecutors and juries in the state consider to be core capital murders—cases in which the evidence is strong enough and the offense aggravated enough that death nearly always is imposed—and to distinguish "outlier" verdicts that are imposed for offenses the state's prosecutors and juries do not consistently treat as warranting a death verdict. Rigorously identifying core capital cases and reversing exceptional uses of the penalty narrows the risk of error identified by our analyses in three ways:

• Conscientious comparative review derives community death-sentencing standards from the strong trend over time of charging decisions by the state's prosecutors and guilt-innocence and sentencing decisions by the state's juries. Across cases and over time, those decisions can more narrowly identify the quality of evidence and level of aggravation that is generally thought to warrant capital treatment in the state.

• Comparative review can moderate high death-sentencing rates, and the high error rates associated with them, by quickly screening out the weak cases in which serious error is most likely and by advising prosecutors and jurors against future sentences in similar cases.

• The standards the court sets give prosecutors and trial judges a mandate and a means to resist political pressures to over-charge and to conform their decisions to popular sentiment, further lowering the risk of error that those pressures otherwise create.

The six options discussed above are policies and standards designed to focus capital charges and verdicts on classes of cases in which the evidence of guilt and the amount of aggravation net of mitigation is clear and strong, and to exclude marginal categories of cases in which the risk of error is high. These reforms would:

• moderate high capital-sentencing rates that are associated with high rates of serious error;

• help shield prosecutors and judges from political and race-related pressures that lead to over-use of the death penalty and resulting error and risk;

• increase attention paid to—and in some cases directly improve the quality of—the jurisdiction's non-capital response to crime, the ineffectiveness of which is also related to high rates of capital error; and

• focus on capital policy—the guidelines defining the broad categories of cases in which the state or county invites capital charges, convictions and sentences, which are the kinds of official activities our analyses most strongly link to capital error.

As is noted above, however, the political pressures to overuse the death penalty that our analyses link to high capital error rates may keep policy makers from adopting these changes in an effective form or from sustaining them once they are adopted⁹²¹—allowing chronically high error rates to persist or to reappear, and keeping the system from achieving its goals. The four options set out below take a different approach. Instead of placing entire sets of marginal cases off limits to capital outcomes, these proposals aim to improve the quality of decisions in each case by prosecutors, judges, jurors and defense lawyers, so that they can more reliably separate marginal from core candidates for capital verdicts.

Each of the options described below aims to increase the capacity of particular actors in the death penalty process to serve as a check on excessive capital-charging and capital-sentencing policies and practices that our analyses so strongly associate with a high risk of serious capital error. Some of these options might make capital trials last a few days longer and cost more. Our study findings reveal, however, that increased funding at the front-end of the capital process will more than pay for itself through reduced costs at the back-end of the process and the quelling of doubts about the integrity of the system. This point was succinctly made recently by Indiana University Law Processor Henry C. Karlson, "a staunch supporter of capital punishment" and frequent advisor to state lawyers defending capital cases: "We spend very little money on trial, then spend a great deal on appeals," Karlson said. "That's idiocy. I say do it right the first time and you won't need 20 years to figure out if anything went wrong."⁹²²

g. Base charging decisions in potentially capital cases on full and informed deliberations.

As various observers have concluded, spur-of-the-moment capital-charging decisions by single prosecutors before all the evidence is available can often commit jurisdictions to capital prosecutions in cases where the evidence of guilt is not strong, or evident aggravating circumstances are substantially offset by later-discovered extenuating factors.⁹²³ Once a case is charged capitally, substituting non-capital charges, or the jury's imposition of a non-capital conviction or sentence, may be seen as a defeat for law enforcement—even when that outcome is the appropriate one, given the evidence, circumstances and law. Over-charging of this sort in turn puts strong pressures on officials to cut corners and overstep bounds to avoid defeat, and to secure a capital conviction and sentence notwithstanding the weak evidence and aggravation or the strong mitigation.⁹²⁴

Capital statutes adopted recently by Congress and New York, and local practices in places like Austin, Texas, and Jacksonville, Florida,⁹²⁵ have identified a useful way to limit this problem. In those jurisdictions, a decision to proceed capitally in cases in which murder charges have been filed may not be made until three things have occurred:

• police and prosecutors have completed their own investigation into the offense, to determine the strength of the evidence of guilt and the balance of aggravating and mitigating factors;

• defense lawyers, following their own investigation, have been invited to meet with prosecutors, review the state's and their own evidence, and explain why capital charges are not warranted; and

• multiple individuals associated with the prosecuting office—some responsible for investigating the case, others not, some able to compare the case to similar situations where final verdicts are known, and all aiming to identify and follow local standards for limiting capital charges to "the worst of the worst"—conclude, based on all the evidence and information, that capital charges are warranted.

These proposals track the advice of Joe Birkett, President of the Association of Government Attorneys in Capital Litigation and a Chicago-area prosecutor. Birkett reportedly recommended (1) that prosecutors "[e]liminate knee-jerk [charging] decisions" by using "written policies for deciding whether to seek the death penalty in murder cases" and "capital-case committees," or by "consult[ing] with more experienced prosecutors elsewhere," and (2) that "[b]efore deciding whether to seek the death penalty, prosecutors should [invite] defense attorneys to submit mitigation packets—information on a defendant's mental state and upbringing that could evoke sympathy at trial."⁹²⁶ In keeping with our study findings, these steps can:

• replace the indiscriminate capital-charging policies that our study strongly associates with increases in capital error;

• target capital charges on the strongest cases for a death verdict where serious error is least likely;

• help local professionals use their own standards—and any standards in state statutes or developed during proportionality review by state high courts—to resist the pressures to over-use the death penalty that also are related to high error rates; and

• foster improved law-enforcement, which is associated with lower capital error rates.

h. Make all police and prosecution evidence bearing on guilt vs. innocence, and aggravation vs. mitigation, available for presentation at trial.

The best single source of information on the strength of the evidence of guilt and the amount of aggravation net of mitigation in any given case is the police and prosecution file in the case.⁹²⁷ Often, however, potentially important evidence in the file never reaches the jury:

• The failure of police and prosecutors to disclose evidence of innocence and mitigation is the second or third leading reason state post-conviction and federal habeas judges overturn capital verdicts.⁹²⁸

• The failure of police and prosecutors to disclose evidence before trial is one of the main reasons post-trial litigation over the reliability of capital verdicts takes so long—now averaging 12 years from death sentence to execution.⁹²⁹

• Prosecutors' charging decisions are more likely to ignore the weakness of the evidence of guilt and aggravation, and the strength of mitigation, when the rules allow them to shield evidence of problems with their case from public view.

One reason official suppression of important evidence is common before trial, and the subject of such costly and contentious litigation on appeal, is that the legal rule stating when police and prosecutors must turn over evidence is ambiguous and difficult to apply. Under that rule, the decision whether a police officer or prosecutor must turn over evidence indicating that the defendant may be innocent turns on a police officer's or a prosecutor's guess about how the evidence might or might not change events at a trial that has not yet occurred.⁹³⁰ To avoid the problems such guesswork creates, a number of capital prosecutors around the country—including most federal capital prosecutors—follow an "open files" policy making all the evidence in their and law enforcement files available to defense lawyers, who then decide whether there is anything in the file to present to the jury as evidence of the weakness of the state's case or in support of a defense.

Our analyses reveal that it is in close cases—those in which a small amount of evidence might tip the outcome in a different direction—that the risk of serious error is the greatest.⁹³¹ And yet under existing rules, it is in just those cases that officials are especially likely to conclude that disclosure of the seemingly small amounts of exculpatory evidence in their files is not required because the officials do not believe (as existing rules require before disclosure is mandatory) that the evidence would probably change the outcome of trial.⁹³² Our findings suggest that any jurisdiction that relies on fully informed and responsible capital juries or judges to identify the "worst of the worst" cases, and to screen out the rest at the conviction and sentencing stages, and yet that does not insist that those decisions be informed by all the available evidence, takes an intolerable risk of over-conviction, over-sentencing and serious error, including execution of the innocent. Because doing so also discourages prosecutors from making hard-headed evaluations of the true strength of the evidence of guilt and aggravation when they charge cases capitally, keeps defense lawyers from doing their jobs properly at trial, and lengthens appeals, open-files policies in capital cases are, in our view, a second of three policy imperatives.⁹³³

i. Insulate capital-sentencing and reviewing judges from political pressure.

Our analyses show a clear and consistent connection between political pressures elections put on state judges and higher rates of serious capital error.⁹³⁴ In making this finding, we rated states based on a variety of selection techniques, each of which places additional political pressure on state judges and all of which together are associated with higher capital error rates. Other findings also reveal political pressure on elected state appellate judges to affirm seriously flawed death verdicts.⁹³⁵ Given these findings, each change listed below would decrease the risk of serious trial error and increase the likelihood that state appellate judges will correct such error when it occurs:

• Appoint rather than elect capital trial and appeals judges.

• Lengthen those judges' terms, whether they are appointed or elected.

• If judges are elected, use non-partisan elections or recall or retention elections, rather than contested elections.

• If states do not wish to give up frequent, partisan election of judges in order to improve the reliability of capital verdicts, they should use jurors to determine the sentence in capital cases—a practice the Supreme Court has recently suggested might be constitutionally required.⁹³⁶

j. Identify, appoint and compensate capital defense counsel in ways that attract an adequate number of well-qualified lawyers to do the work.

Our adversarial system relies almost exclusively on a single, case-by-case check on false and over-broad criminal charges, convictions and sentences—the diligent testing of the state's case by a lawyer for the accused who is as skilled, well-funded and determined as the prosecution. If this check functions properly, it can help alleviate many of the conditions our analyses show are most strongly associated with serious capital error:

• One of the most effective checks against convicting defendants who are innocent of a capital offense is a skilled and determined criminal defense lawyer with the financial support and investigative and expert help needed to reveal defects in the state's case and demonstrate the client's innocence.

• As is documented by a careful study of improved standards for appointing, compensating and providing support services to capital defense lawyers in Indiana and by preliminary data on Oklahoma, the routine provision of qualified and adequately funded capital trial lawyers leads to sharp decreases in death-sentencing rates. And our findings link decreases in death sentencing to substantially lower capital error rates.⁹³⁷

• As the Indiana study also documents, an assurance of qualified counsel dissuades prosecutors from bringing capital charges in weak cases, given the likelihood that the lawyer will identify weaknesses in the state's case, convince jurors to forgo convicting when the state's case is doubtful, demonstrate the inappropriateness of the death penalty for defendants with inherently extenuating conditions, and identify alternatives to a death sentence that jurors find sufficient to punish the defendant and protect society.

• In these ways, qualified and adequately compensated counsel help assure that capital convictions and death sentences are only imposed when the probability of serious error is the lowest: i.e., for highly aggravated killings where the evidence in support of the capital verdict is strong.

• Skilled lawyers are also more likely to

  • insist that the states provide the necessary resources for investigations, DNA and other forensic analyses, psychiatric evaluations and other support services that are required to avoid serious error;

  • dissuade politically vulnerable judges from making the kinds of erroneous rulings that political, racial and other pressures otherwise tend to trigger; and

  • expose the weaknesses in law enforcement strategies that are associated with high rates of capital error—and expose the insufficiency of aggressive capital charging and sentencing as a stop-gap response to such weaknesses.

This crucial adversarial check on flawed capital trials that is provided by competent and adequately funded defense lawyers has broken down in many capital jurisdictions:

• Egregiously incompetent lawyering—the only kind for which reversal is permitted⁹³⁸—is responsible for about 40% of reversals at the state post-conviction phase of capital review and between a quarter and a third of the reversals at the federal habeas stage.⁹³⁹

• An important predictor of high error rates at the state direct appeal review stage is poorly funded courts.⁹⁴⁰ Inadequately funded state courts are in turn associated with poorly funded capital defense lawyers, who typically are compensated out of court budgets. Moreover, states that resist spending money on their criminal courts are also likely to skimp on criminal defense.

• The literature abounds with horror stories about the low quality of counsel in capital cases.⁹⁴¹

• Observers also have frequently linked low-quality lawyering both to higher rates of capital-sentencing⁹⁴² and to higher rates of serious error including the conviction and capital sentencing of the innocent.⁹⁴³

The main cause of the breakdown of the adversarial check in capital cases is a dangerous combination of very heavy demands on capital defense lawyers and very low compensation. The amounts of lawyer time and expert and investigative resources needed for an adequate defense in a capital cases are high—many times those needed for the typical non-capital defense.⁹⁴⁴ Most estimates of the cost of a minimally adequate capital defense, or of the market rates that prevail for such cases, run from $50,000 at the low end to $250,000 or more in urban areas.⁹⁴⁵ Yet the resources states make available to compensate capital lawyers and defray their expenses are often less than 10% of the going rate for a minimal defense—and frequently are pegged to the level deemed minimally adequate for working out a plea bargain in, for example, a common larceny case.⁹⁴⁶ Some states cap lawyers' compensation for an entire capital trial at $5000, or even $1000.⁹⁴⁷ Other states put ceilings of $1000 to $5000 on funds available for investigators and expert assistance, even though an adequate defense, particularly in factually complex cases and ones with mental health issues, requires tens of thousands of dollars in support services.⁹⁴⁸

As Fordham University Law Professor Bruce Green points out, "[t]he paltry compensation provided to lawyers who are appointed to defend capital cases . . . discourages members of the private bar from developing an expertise in death penalty litigation."⁹⁴⁹ As University of North Carolina Professor James Coleman has further documented, the heavy time demands that conscientious lawyers feel compelled to meet if they accept appointments in capital cases in return for minuscule compensation and reimbursement of expenses "almost inevitably mean that virtually the only lawyers who are willing to handle capital cases are inexperienced, ill-prepared and under-funded."⁹⁵⁰ The result is that capital defendants are routinely represented by lawyers:

• with only a few years or months at the bar, no qualifications besides political-patronage connections to the judge who appointed them or a willingness to speed trials along to an inevitable conclusion,⁹⁵¹ histories of bar discipline,⁹⁵² drug and alcohol problems,⁹⁵³ difficulty staying awake during trial,⁹⁵⁴ and either no capital or even criminal law experience or (at the other extreme) back-breaking criminal caseloads;⁹⁵⁵

• who fail to conduct the necessary factual investigations and legal research,⁹⁵⁶ never seek expert assistance, forgo obviously valid motions and objections, and overlook readily available and important evidence;⁹⁵⁷ and

• who as a result are no match for the "prosecution [whose] resources in death penalty cases," according to a Coopers & Lybrand study of the costs of capital representation, often "seem[] unlimited."⁹⁵⁸

As Northwestern University Law Professor Larry Marshall has concluded, "If a criminal trial is to be a fair search for the truth, then each side must be given relatively equal resources with which to investigate and present its version of the truth. This is almost never the case with capital trials."⁹⁵⁹

Scores of state and federal court reversals summarized in Appendix C and D document these and other break-downs in the adversarial check on inaccurate and unreliable capital convictions. An example is the federal habeas reversal of Frederico Martinez Macias's capital verdict. Macias was capitally charged with killing a man during a burglary in El Paso, Texas in 1983. Macias was implicated in the murders by a co-worker in whose yard some of the victims' property was recovered. Although the co-worker was an ex-convict who admitted taking part in the burglary-murder and disposing of the stolen property, he escaped prosecution in return for testifying against Macias. Other testimony came from jailhouse informants who said Macias confessed the crime to them, and from a nine-year-old witness who claimed to have seen Macias washing blood off his hands. A jury convicted Macias and sentenced him to die. Following direct appeal, Macias came within two days of being executed before a stay of execution was granted and lawyers at the Washington, D.C. law firm of Skadden, Arps, Slate, Meagher & Flom agreed to represent him on a volunteer basis in 1988. After two more years of intensive investigations and state and federal court review, during parts of which the Skadden lawyers were assisted by an Oklahoma Law School professor, a federal district judge found that Macias's trial attorney had provided ineffective assistance of counsel at trial by failing to (1) conduct even a cursory investigation, (2) identify and call two alibi witnesses, (3) call eyewitnesses who had seen the co-worker and another suspicious man—not Macias—at the victim's home, (4) challenge the 9-year-old's questionable memory, (5) present other witnesses who could have rebutted the prosecutor's case, and (6) conduct the rudimentary legal research needed to avoid a damaging trial mistake on a key evidentiary point. According to the district judge, "[t]he errors that occurred in this case are inherent in a system which paid attorneys such a meager amount."⁹⁶⁰ In affirming the reversal, a conservative United States Court of Appeals panel

We are left with the firm conviction that Macias was denied his constitutional right to adequate counsel in a capital case in which actual innocence was a close question. The state paid defense counsel $11.84 per hour. Unfortunately, the justice system got only what it paid for. . . . The judgment of conviction is vacated and Federico Martinez-Macias shall be released from custody if the State of Texas has not commenced a new trial within 120 days of our mandate.⁹⁶¹

Macias thereafter was freed when a grand jury found insufficient evidence to re-indict him.

Macias's case illustrates the high costs to the system of poorly qualified and compensated capital trial counsel. Vastly offsetting the resources saved by funding only a short trial and a single lawyer at $12 an hour are the costs to Macias himself of spending nearly a decade behind bars despite the absence of evidence against him; to the family of the murder victim, which waited the same amount of time for the adjudicated killer to be punished, only to learn that there was no reliable evidence against him and that others who probably committed the crime had never been arrested; to 10 state and federal judges who devoted many hours and resources to the case; to the army of lawyers who eventually came to Macias's rescue; and to the public, which paid for those appeals and lawyers and whose safety to this day is imperiled by the prosecution's failure to discover the weakness of its case and bring the real culprit to justice.⁹⁶²

In jurisdictions relying on a case-by-case response to the problem of serious capital error and the consequent breakdown of the capital system, it is imperative that such responses include:

• standards assuring that only well-qualified lawyers represent capital defendants;

• methods of appointing capital lawyers that avoid patronage considerations and rewards to financial contributors to judicial campaigns; and

• sufficient compensation and reimbursement for experts, investigators and other litigation necessities to trigger the formation of a stable and qualified capital defense bar.

C. Changes Likely to Magnify the Problem of Serious Capital Error

In addition to suggesting reforms that can help alleviate chronically high rates of serious capital error, our study results identify changes in existing practice that will not likely reduce error, and may well make things worse. Four unproductive approaches are discussed below.

1. Doing nothing is not an effective response to chronically high error rates and may well let them get worse.

The death penalty system is broken. And forces are at work that seem to be making it worse. At the state direct appeal review stage—the only stage that reviews all capital verdicts, and the stage responsible for nearly 80% of all reversals during the 23-year study period—more recent verdicts were significantly more likely to be found seriously flawed than earlier verdicts after controlling for other factors.⁹⁶³ There is no reliable evidence that the problem of chronically high error rates will resolve itself over time, absent meaningful reforms.

2. Cutting back on review of capital verdicts may increase the ill-effects of chronic error and lead to more error.

As we note earlier, the main changes in capital practice since the study period ended are sharp cutbacks in the breadth of appellate review in places like Texas at the state post-conviction stage, and nationwide at the federal habeas stage.⁹⁶⁴ But, as New Mexico Governor and long-time death penalty supporter Gary Johnson recently said in withdrawing a proposal to limit capital appeals, this kill-the-messenger strategy is an invitation to disaster—a change, in Johnson's words, that could "lead to innocent people being executed."⁹⁶⁵ Although as we have seen, the review process does not effectively feed back the information needed to improve capital trials, often misses serious errors, and performs unevenly as a substitute sentencer,⁹⁶⁶ the review process nonetheless has come to serve a crucial role in screening out large numbers of unreliable death verdicts. Indeed, appeals screened out more than 7 seriously flawed verdicts for every execution during the study period. Especially if capital error rates continue to occur at anything like the rates during the 23-year study period, the effect of limiting inspections for error almost inevitably will be to decrease the probability that serious errors will be corrected, and to increase the risk that innocent inmates will be executed. Limiting inspections can even cause error rates to rise, by removing the only existing, if weak, deterrent to the conditions associated with error, such as capital over-charging and over-conviction, political and race-related pressures on capital officials to expand capital punishment in lieu of effective law enforcement strategies, political pressure on trial judges to tailor rulings to popular sentiment, and under-funding of state criminal courts.

3. Piecemeal additions to the list of qualifying aggravating circumstances may increase capital error rates.

Another common modification of capital statutes is to add new aggravating factors that allow the imposition of the death penalty. The temptation to do this is great in the aftermath of a heinous crime to which no existing aggravating factor applies. But long lists of statutory aggravating factors—an attribute, for example, of Illinois's capital statute that some have linked to its high rates of serious capital error—establish just the kind of broad death-sentencing policy that are associated with high error rates.⁹⁶⁷ New aggravating circumstances should be resisted on principle. If consideration is given to them, any change should not proceed piecemeal, but should be part of an overall revamping of the statute that removes other, less appropriate circumstances at the same time. This option again tracks the advice of Joe Birkett, President of the Association of Government Attorneys in Capital Litigation, that "[m]any states list too many aggravators" in their statute and should "streamline" those lists.⁹⁶⁸ Prime candidates for removal, given their tendency to vastly expand the reach of capital statutes and to sweep in cases that are not highly aggravated are:

• catch-all aggravating circumstances that are vague and apply to essentially all first-degree murders—e.g., that the offense was "especially heinous," "atrocious," "horrible" or "depraved," or "above the norm of first-degree murder";

• aggravating factors that simply duplicate the definition of murder and thus do nothing to assure that killings for which the death penalty is available are more aggravated than most murders—e.g, that a killing occurred in the course of a felony or was premeditated, which are prerequisites to virtually all first-degree murders; and

• repetitious aggravating factors that treat the same fact as two different reasons to impose death, inviting prosecutors and sentencing juries to inaccurately inflate the seriousness of the offense by double-counting a single aggravating trait—e.g., that the murder was both "in the course of a robbery" and "for pecuniary gain," or "the victim was a police officer" and the killing "avoided lawful arrest."

4. Large-scale underwriting by the state of the costs of local capital prosecutions invites higher capital error rates.

The final category of reforms against which our results caution are ones that largely replace local with state financing of capital prosecutions. These kinds of reforms may have the best of intentions—improving the quality of capital prosecutions, or defraying retrial costs so localities can avoid errors made at the first trial. But our analyses strongly link capital error to policies that reward local officials who use the death penalty broadly, while displacing the post-trial cost of the errors these policies trigger onto taxpayers across the state and nation.⁹⁶⁹ If, in addition, local officials and their constituents can avoid most of the costs of the initial trial—and, worse, of the retrials their errors require—they will have even less reason to use the penalty judiciously. Open-ended state subsidies also make it easier for local officials to give in to political and race-related pressures to use the penalty broadly—including to mask failings of the county's other law enforcement strategies—amplifying capital error still further.