When Mercy Discriminates

Article - Volume 102 - Issue 7: Mercy Symposium

“[T]he power to be lenient is the power to discriminate.” – Kenneth Culp Davis[1]

The Supreme Court cited the Kenneth Culp Davis quote above when it rejected a challenge to Georgia’s capital punishment scheme as racially discriminatory in McCleskey v. Kemp.[2] McCleskey provided the Court with powerful evidence of racial bias in Georgia’s administration of the death penalty.[3] Specifically, McCleskey relied on a study by David Baldus and his colleagues finding that a defendant who killed a white victim was 4.3 times more likely to get the death penalty than a similarly situated defendant who killed a Black victim.[4] The Court assumed the validity of Baldus’s findings but still ruled against McCleskey because it accepted disproportionate results as the price to be paid for allowing juries to spare some defendants a death sentence.[5] As the opinion by Justice Powell put it, “a capital punishment system that did not allow for discretionary acts of leniency ‘would be totally alien to our notions of criminal justice.’”[6] The potential for racial bias, the Court seemed to be saying, is the price we have to pay for leniency.

Racial bias is a high price, indeed, and not everyone has been willing to pay that price. The most common response to unequal treatment on the basis of race has been to curb the discretion that permits leniency.[7] That was the rationale behind mandatory sentencing laws, which sought to prevent judges from granting leniency to well-off defendants who were disproportionately white.[8] This type of solution to the mercy/disparity conundrum assumes that equality concerns outweigh the value of leniency.

There are two flaws with making a tradeoff of leniency for a promise of equality. First, this option undervalues the need for leniency, particularly in America, where so many forces push the law toward unjust severity. The push for equality under this model gives more people severe punishments rather than finding ways to give leniency to a wider group.[9] Second, these strategies have failed to foster equality, so they have been false trades in which leniency was relinquished for nothing. That is because discretion is never fully eradicated. It tends merely to shift form, and the attempts to mandate punishment have resulted in the same disparate outcomes they were designed to remedy. Indeed, often the outcomes have been even more discriminatory under models that have attempted to stifle discretionary leniency. That has been the central story of mandatory minimums and sentencing guidelines, the two most prominent efforts at addressing racial disparities in the exercise of leniency. Instead of trading leniency for equality, we have often been left with more severe sentences and the same or worse disparities.

Although they have received less attention, there have been better strategies for addressing unwarranted racial disparities while preserving leniency. In both the dispensation of clemency and compassionate release, we have seen successful examples of leniency that have not produced racially disparate outcomes. These models focus on broad categories of eligibility that are sensitive to avoiding unwarranted racial disparities. By using this kind of categorical approach transparently, we can reduce disparities while preserving the opportunities for leniency that are more necessary than ever in an excessively punitive system.

This Essay analyzes the relationship between mercy and discrimination in the administration of criminal law and punishment in three parts. Part I begins by describing the relationship between leniency and discrimination. Part II explores what has been the most popular approach for tackling this bias, which is the use of substantive limits on when mercy can be given. Part III describes what more effective strategies would look like.

I. Biased Mercy

A decision is an exercise of mercy whenever an official has the legal grounds and authority to impose a punishment on an individual or to leave a punishment already imposed in place but opts not to because of sympathy or compassion.[10] Not every reduction in a sentence is properly thought of as merciful. For example, it would not be an exercise of mercy to reduce or withhold punishment for the personal convenience of the decision-maker or the desire to conserve public resources. While the recipient might experience the benefit as merciful, if the decision-maker did not reduce the punishment out of some sympathy for the defendant, it is not really an act of mercy.

Officials have discretion to dispense mercy across the range of decisions in the administration of criminal law and punishment.[11] Police officers could elect not to stop or arrest someone. Prosecutors could decide to dismiss or lower charges. Juries could acquit or, in capital cases, vote not to impose a death sentence. Judges could give lesser punishments or approve motions for compassionate release. Parole boards could approve an earlier release. Executives could grant clemency. Across all of these domains, there are racial disparities in who benefits, with white people being far more likely to be on the receiving end of leniency.[12]

In some of these contexts, it would be hard to know that a decision to be lenient is based on mercy as opposed to some other factor. When the police opt not to stop or arrest someone, for instance, we lack reliable data as to the reasons. It could be the officers are giving someone a second chance because they are sympathetic to the individual’s situation, or it could instead be that the officers do not want to bother with paperwork. Likewise, when prosecutors offer defendants lesser sentences if they plead guilty, it is possible they are showing mercy. However, it is equally plausible—likely more so—that the reduction is offered for administrative efficiency because the prosecutor wants to avoid going to trial. It is, of course, still noteworthy and disturbing that collectively police officers and prosecutors are making these decisions on a racially skewed basis. But it is hard to pinpoint these choices as exclusively based on who should get mercy.

In other contexts, however, we can have greater confidence that a decision toward leniency is grounded in the traditional conception of mercy that I defined above. Clemency decisions—whether to give pardons or commutations—are the prototypical examples. While a governor’s or a president’s decision to deny a clemency request is typically the product of political considerations, the decision to grant clemency is rarely based on some expected personal or public benefit for doing so. Quite the opposite, officials usually reduce sentences or pardon convictions at great risk to their political future. Their decisions in favor of leniency are instead typically driven by the facts of an individual’s case and the emotional connection they feel. For example, former Arkansas Governor Mike Huckabee faced political blowback in his run for president when an individual who received a sentence commutation from him later killed four police officers. One of his staffers on his presidential campaign noted Huckabee’s clemency decisions emanated from his “empathy for all people and genuine belief in the individual” but that “[t]he unfortunate reality is that for politicians, unlike pastors, there are limits to compassion.”[13] Former Mississippi Governor Haley Barbour, who was criticized for giving clemency to more than two hundred people at the end of his term in office, responded that he “believe[s] in second chances, and [he] tr[ies] hard to be forgiving.”[14] Current Minnesota Governor Tim Walz defended the Minnesota clemency process as a place “where we believe people can change, where we believe that we can have healing, where we believe we can have restorative justice.”[15]

When judges give lesser sentences or juries opt against capital punishment, those exercises of leniency are also often grounded in considerations of mercy. The judges and juries have nothing personal to gain from being lenient. They are instead focused on the person before them and the facts of the case.[16]

Clemency and sentencing are therefore two contexts that offer a valuable window into who gets the benefit of mercy—who is garnering sympathy and understanding and forgiveness—when decision-makers have discretion. The evidence of racial disparities in both contexts is clear.

Let’s start with executive clemency. While we could use better data on who is applying for clemency to better analyze grant rates in terms of the total applicant pool, the available information shows that women are more likely to get clemency than men and that white clemency seekers are generally more likely to get relief than Black people and other people of color seeking clemency.[17] The most recent comprehensive study looked at data from thirty-nine states and found “a significant and troubling racial gap in grants,” with non-white applicants being far less likely to receive relief.[18] Studies of clemency grants in particular jurisdictions have mirrored these findings. For example, of the 221 people granted clemency by Governor Haley Barbour in Mississippi, 64% were white and 31% were Black, even though Black people made up 64.9% of the state’s prison population at the time, and white people made up only 33.9%.[19] A ProPublica investigation of clemency during George W. Bush’s presidency found that all but thirteen of the 189 people pardoned by President Bush were white, including all thirty-four of the people pardoned for drug offenses.[20] The grant rate during Donald Trump’s presidency were similarly skewed, as only 18% of his grantees were Black, even though Black people comprised 38% of the federal prison population.[21]

Sentencing decisions also differ based on a defendant’s race, with many studies finding that white defendants are more likely to get more lenient treatment. There is ample evidence of racial disparities in the decision whether to incarcerate someone for a crime or instead to give some other kind of sanction such as probation (what is often known as the “in-out decision”).[22] Studies also show racial bias in sentence length, with white people receiving shorter terms of incarceration than people of color.[23] Women also tend to receive more lenient treatment than men.[24]

Similar evidence of racial disparities has been found in capital cases. Subsequent research has corroborated the Baldus study findings that defendants are less likely to get the death penalty in cases involving Black victims than white victims. A study looking at 14,749 North Carolina homicide cases from 1980 to 2007 found that the chances of being sentenced to death were approximately three times greater for defendants who were convicted of killing a white victim than for defendants who were convicted of killing a Black victim.[25] John Donohue conducted a similar study in Connecticut, looking at the application of the death penalty in 4,686 murder cases tried in Connecticut from 1973 to 2007.[26] After analyzing the data, he found that “minority defendants who commit capital-eligible murders of white victims are six times as likely to receive a death sentence as minority defendants who commit capital-eligible murders of minority victims (12 percent versus 2 percent)” and that “[m]inority defendants who murder white victims are three times as likely to receive a death sentence as white defendants who murder white victims (12 percent versus 4 percent).”[27] Researchers have also shown that, among those who kill white victims, people with more stereotypical Black facial features were more likely to be sentenced to death.[28]

There is therefore no denying that decisions to give mercy in the context of clemency and sentencing have a racial skew. There is more mercy for defendants who are white, and in capital cases, there is a greater willingness to punish people less harshly if their victims are Black. No one should receive more or less punishment because of their race or the race of their victim, so the question is how best to address this disparity.

II. The Myth of Mandated Equality

The most popular response to evidence of racial disparities in the exercise of leniency has been to curb the discretion that allows leniency to operate in the first place. This is why mandatory minimum sentencing laws and binding sentencing guidelines proliferated in the 1980s and 1990s.[29] By the end of the 20th century, “every state and the federal government [had] some type of mandatory sentencing law.”[30] While those on the right wanted to see more severe laws, those on the political left supported these same reforms because they wanted to address glaring racial disparities in sentencing.[31] The hope was that limiting judicial discretion would bring greater equality.[32]

In reality, those on the right got what they wanted in terms of more severity, but those on the left did not achieve their goal of greater equality. In fact, these binding laws resulted in huge racial disparities.[33] Prosecutors did not bring charges carrying mandatory minimum sentences proportionately across all racial groups but instead were far more likely to bring a charge carrying a mandatory minimum sentence against Black defendants.[34] The United States Sentencing Commission has concluded that sentencing guidelines and mandatory minimum statutes together have had “a greater adverse impact on Black offenders than did the factors taken into account by judges in the discretionary system in place immediately prior to guidelines implementation.”[35]

It is harder for legislatures to curb the discretion associated with clemency than with sentencing because clemency powers are often vested in a chief executive as a matter of constitutional law and cannot be modified by statute. That is true of the federal clemency power, and it is true of many states as well.[36] Even with that high bar, some states have passed constitutional amendments to limit the governor’s discretion to grant clemency. In Pennsylvania, for example, the state amended its constitution to change the requirements for granting clemency to people serving life sentences.[37] The shift happened after an individual who was sentenced to life imprisonment as a juvenile for a homicide received a commutation and went on to commit a series of heinous, high-profile violent crimes.[38] The clemency board that makes recommendations to the governor voted 4–1 in his case in favor of clemency, so the constitutional amendment changed the rule to require unanimous approval by a board before anyone serving a life sentence could receive a commutation.[39]

States have found ways to curb discretion and limit clemency without constitutional changes. In Connecticut, for instance, the state’s Board of Pardons and Parole had changed its commutation policies in 2021 to relax the standards for granting relief and make it easier to remedy excessively long sentences.[40] That policy change resulted in a sharp increase in the number of commutations granted, and nearly two-thirds of those granted relief after this rule change were Black.[41] State Republican lawmakers protested the change in policy, and the Democratic governor caved to that pressure by replacing the head of the board.[42] The new leader of the board then changed the standards for granting commutations to require “exceptional and compelling circumstances” to get a reduction.[43] Predictably, commutations slowed to a trickle as a result, meaning clemency no longer provides a needed corrective for the many excessive sentences handed down by the state that were disproportionately imposed on people of color.

While the sentencing-reform efforts in favor of mandatory minimums and binding sentencing guidelines were propelled by concerns for equality in addition to demands for more severity, the efforts to reduce discretion in clemency grants have been almost exclusively driven by those who believe the harsher the sentence, the better. In both contexts, however, the outcome from reducing discretion has been the same: more severe sentences and huge racial disparities. Curbing the discretion to be lenient has not been a recipe for greater racial equality.

III. Models of Equitable Mercy

The good news is that it is possible both to maintain avenues for leniency and to guard against racially disparate outcomes in the exercise of that leniency.[44] A key starting point is to closely analyze the people and categories that get priority in clemency grants. For example, a RAND Corporation study of pardons from 2001 to 2012 found that factors correlated with race may account for differential grant rates.[45] These include the nature of the underlying offense (with white collar offenders more likely to receive a grant), steady employment, charitable activity, assistance of legal counsel, and positive recommendations from the probation office.[46] Some of these factors, like charitable giving and legal assistance with a clemency application, are associated with greater wealth, and structural inequality means that these factors will inevitably disfavor people of color. Structural bias in American society and the relative lack of employment opportunities in poorer minority neighborhoods means that taking into account steady employment will also lead to racially biased outcomes. A positive recommendation from probation officers might itself be the product of bias because it is unclear how those officers are deciding which applications to support. A preference for granting clemency in white collar crime cases will also lead to disparities because of the racial makeup of that offending population.

Identifying the factors driving decisions thus helps to point the way toward a solution. If legal assistance is making a difference, the solution might be to set up pro bono assistance for all those seeking clemency. If other factors correlated with race are driving biased outcomes, the solution might be to cease having those factors drive decisions or to redefine the categories so that they are more inclusive while still achieving the same aims. If, for instance, employment is used because it is associated with lower recidivism, one could look for other factors that are tied to lower recidivism but do not have the same built-in bias. For instance, recidivism risks are lower as people age, so prioritizing elderly applicants can both keep recidivism concerns in check while also minimizing racial disparities. It is beyond the scope of this Essay to identify which factors should be given primacy in a clemency determination. Rather, the point is to note how important it is to understand how the selection of those factors can be driving disparities in unnecessary ways.

President Obama’s clemency initiative offers an example of a clemency process that took this kind of approach and led to clemency grants that more closely mirrored the actual population of those in prison. When President Obama announced his clemency initiative in the second term of his presidency, he urged attorneys in the Pardon Office of the Department of Justice to focus on people who would not have received the same sentence today because of changes in the law.[47] A large proportion of those serving sentences that would not be legal today were people convicted of crack offenses, and Black people made up the vast majority of those convicted of federal crack offenses.[48] Right off the bat, that kind of categorical line helps to frame the decision in a way that will be racially inclusive. That is, a broadly inclusive category that will both focus on a central harm clemency seeks to rectify excessive sentences, and that will encompass a population that mirrors the federal prison population overall and not just skew in favor of the wealthier white applicants who previously succeeded at disproportionate rates in the process.[49]

The same benefits came from another Obama clemency requirement that individuals had already served at least ten years of their sentence. This, too, is a racially diverse group, and it targets those who have already served substantial time, thus making additional time less likely to bring deterrence or incapacitation benefits.[50]

Additionally, President Obama urged lawyers to take clemency cases pro bono, which helped to address the discrepancy that previously existed between those who applied with and without counsel. An army of attorneys took these cases on for free, thus helping every applicant put their applications in the best light.

To be sure, not all of the Administration’s criteria were racially inclusive. Some were more likely to limit eligibility in racially disparate ways. For example, by specifying that people with “a significant criminal history” should not receive clemency,[51] the Administration was going to disadvantage people subject to the highest levels of policing, which in American has invariably meant communities of color. The initiative could have taken a more nuanced approach to criminal history that accounts for disparate policing while still capturing its central concern of not granting clemency to individuals who pose a greater risk of future offending given their past practices. For example, we know that the policing of drug-related charges disproportionately targets Black and Hispanic people because the police are making discretionary decisions to target specific areas and people to stop. A better approach would therefore make people ineligible for consideration only if they had a significant criminal history made up of the kinds of offenses that are not based on police discretion, such as crimes involving identifiable victims. This could allow a more individualized approach that does not automatically rule someone out because they are likely to receive more police attention.

The factor ruling out individuals who have significant ties to gangs was likewise flawed. Assessments about gang affiliation are notoriously biased to include young men of color without much or sometimes any proof that they are part of a formal gang as opposed to just being in a group of friends.[52] This factor is therefore likely to inject bias in the decision-making. A better approach would have ruled out only those people with a record of leadership or evidence of active participation in a gang and not just used the vague “significant ties” language.

Again, the point is not to decide exactly which criteria should be used but to see how investigating a factor’s link with racial disparities can lead to a better articulation of who should be deemed in or out of consideration as a categorical matter.

A chief lesson from the Obama initiative is that, because the overarching category targeted for relief was itself predominantly made up of Black and Hispanic people, the various limits that could have produced racial disparities did not end up doing so because the overall pool being targeted was itself so diverse. President Obama’s clemency recipients were 70.9% Black, 19.1% white, and 8.7% Hispanic.[53]

Federal compassionate release provides another example of how creating broad and racially inclusive categories for relief can help address concerns with racial disparity in grants of leniency. Individuals can obtain compassionate release based on having specified medical circumstances; being sixty-five years and older with deteriorating health; having family circumstances where the defendant is the only one available to provide needed care to a minor child, spouse, or parent; being the victim of abuse while in custody; or serving an unusually long sentence.[54] These categories are racially inclusive and thus establish an eligibility pool that does not disproportionately leave out people of color. It has, accordingly, yielded results that do not suffer from some of the disparities we have seen when categories are not created in this way.[55] Studies of federal compassionate release have found that the grant rate varied by only 2.2% across the different racial categories.[56]

Another valuable tool for checking bias that goes hand-in-hand with creating broad and inclusive categories for leniency is data collection. Oftentimes, decision-makers do not even realize their decisions are biased in favor of specific groups. Keeping track of decisions to watch for biased trends is a valuable first step in correcting them. For example, in the study of the death penalty in Georgia, the Baldus study showed that the disparity appeared in a middle category of cases.[57] The most serious cases did not have disparate outcomes on the basis of race, nor did the least serious.[58] One way to address the disparities in Georgia given this knowledge would thus be to more narrowly tailor capital punishment eligibility to the most serious cases and leave out the middle category. The RAND study highlighting the clemency factors that correlate with racial disparities is similarly valuable in identifying which criteria need to be altered to better achieve the goals of clemency without producing such racially disparate outcomes. When you study what is driving the disparities, you can craft solutions to minimize or eliminate them.

Finally, one last strategy for addressing racial disparities in decisions to be lenient focuses on the decision-makers themselves. The ideal institutional model would have a diverse group of decision-makers to help reduce bias.[59] In some contexts in which a single individual is vested with authority to decide, it might be possible to create an advisory body to help improve decisions. For example, the federal Constitution vests the pardon power in the president alone, and that same model exists in some states where the governor has sole clemency authority. These decision-makers can nevertheless create advisory boards that represent a diverse range of viewpoints and experiences to help create a diverse pool of potential clemency recipients.[60] In the context of sentencing, a sentencing commission or board comprised of diverse individuals could provide guidelines for judges to help guide decisions to reduce disparity. Again, the key would be for this body to be diverse and to create broad categories for leniency crafted with racial inclusivity in mind based on the best available data.[61]

Conclusion

Severity and racial disparities are two of the chief concerns with the operation of criminal punishment in America today. Solutions to these twin problems do not have to be mutually exclusive. While it is not possible to completely eradicate racially biased decisions when actors have discretion to be lenient, it is feasible to keep such bias to a minimum while maintaining the crucial benefits that valves for leniency provide. The key is to closely analyze what factors are driving the decisions to be lenient, to craft eligibility criteria in ways that are as race-neutral as possible while achieving the policy goals of the grants, to continuously track decisions with an eye to disparate trends, and to put in place decision-making structures that can act as a check on individual bias. In other words, it is important to pay attention to the substantive criteria being used, to the process for making the decisions, and the institutional design of the decision-making structure. What has seemed like an inevitable link between leniency and disparity is, in reality, more a product of inattention to the issue. There are avenues for providing needed mercy in America’s imposition of punishment without unwarranted racial disparities, and it should be a priority to pursue them.

  1. . Kenneth Culp Davis, Discretionary Justice 170 (1969).
  2. . 481 U.S. 279, 312 (1987).
  3. . Id. at 286–87.
  4. . Id.
  5. . See id. at 311–13 (holding that “[i]n light of” the safeguards in the jury trial process and “the benefits that discretion provides to criminal defendants,” the study did not demonstrate a “constitutionally significant risk of racial bias affecting the . . . sentencing process”).
  6. . Id. at 312 (quoting Gregg v. Georgia, 482 U.S. 153, 199 n.50 (1976)).
  7. . See James Q. Whitman, Harsh Justice: Criminal Punishment and the Widening Divide between America and Europe 53 (2003) (discussing the appeal of formal equality in America, in the form of sentencing guidelines, to prevent disparate treatment by race by limiting judges’ discretion).
  8. . Id.
  9. . Id. at 193–94.
  10. . See Dan Markel, Against Mercy, 88 Minn. L. Rev. 1421, 1436 (2004) (defining “compassion-based” mercy).
  11. . Rachel E. Barkow, The Ascent of the Administrative State and the Demise of Mercy, 121 Harv. L. Rev. 1332, 1334 n.6 (2008).
  12. . Rachel E. Barkow, The Court of Mass Incarceration (forthcoming) (manuscript ch. 6, at 1–2) (on file with author).
  13. . Mark Z. Barabak & Nicholas Riccardi, Mike Huckabee Defends Freeing Convict Wanted in Washington Police Shootings, L.A. Times (Dec. 1, 2009, 12:00 AM), https://www
    .latimes.com/archives/la-xpm-2009-dec-01-la-na-huckabee1-2009dec01-story.html [https://perma.cc/KS7D-2AA5].
  14. . Maggy Patrick, Haley Barbour Defends Decision to Grant Pardons, Says He Believes in ‘Second Chances, ABC News (Jan. 13, 2012), https://abcnews.go.com/blogs
    /politics/2012/01/haley-barbour-defends-decision-to-grant-pardons-says-he-believes-in-second-chances [https://perma.cc/3XCA-9ZYQ].
  15. . Deena Winter, ‘I Can’t Take Back What I’ve Done, Minn. Reformer (Dec. 22, 2023, 10:20 AM), https://minnesotareformer.com/2023/12/22/i-cant-take-back-what-ive-done/ [https://perma.cc/JJ9N-F286].
  16. . See, e.g., Sentencing Hearing Transcript at 31, United States v. Falterbauer, No. 15-CR-00397 (S.D.N.Y. Feb. 24, 2016), https://acrobat.adobe.com/id/urn:aaid:sc:us:845fe82c-6712-425a-863f-dbe182868c53 [https://perma.cc/BCR9-3CH2] (imposing a sentence of probation and community service instead of imprisonment and informing the defendant the judge was “show[ing] you some mercy”); Sentencing Hearing Transcript at 132, United States v. Whatley, No. 08-CR-13 (N.D. Ga. July 31, 2008), https://acrobat.adobe.com/id/urn:aaid:sc:US:9e832121-2470-4d33-8712-b165a1f925ce [https://perma.cc/CH97-GH58] (giving a sentence below the recommended guideline, and explaining: “And it may have been said that a court should not be merciful, but if the day ever comes when a specific court cannot be merciful, then that court needs to terminate its activities. Mercy is the hallmark of justice.”); Sentencing Hearing Transcript at 75, United States v. Proano, No. 08-CR-20544 (S.D. Fla. Feb. 17, 2010), https://acrobat.adobe.
    com/id/urn:aaid:sc:US:064861aa-ba36-44de-b6bb-70fac9888594 [https://perma.cc/JL9P-KPAK] (“The only reason why I didn’t sentence at the high end is that I believe in the [sic] be merciful, and it is only because of the request for mercy. Otherwise I would have sentenced at the top of the guidelines at 97 months.”).
  17. . Anna Gunderson, Who Deserves Mercy? State Pardons, Commutations, and the Determinants of Clemency 13, 15 (Apr. 15, 2022) (unpublished manuscript), https://preprints.apsanet.org/engage/apigateway/apsa/assets/orp/resource/item/6259b360368ab64723809120/original/who-deserves-mercy-state-pardons-commutations-and-the-determinants-of-clemency.pdf [https://perma.cc/5TGX-C6EX]. The pattern of clemency grants in capital cases is somewhat different. As is true for non-capital cases, women are more likely to get a death sentence commuted than men. Michael Heise, Mercy by the Numbers: An Empirical Analysis of Clemency and Its Structure, 89 Va. L. Rev. 239, 277 (2003) (noting that even after controlling for various factors, women are more likely than men to receive a commutation in capital cases); William Alex Pridemore, An Empirical Examination of Commutations and Executions in Post-Furman Capital Cases, 17 Just. Q. 159, 167, 171 (2000) (examining cases from 1974 to 1995 and finding that eight of the nine women sentenced to death received a commutation); Elizabeth Rapaport, Equality of the Damned: The Execution of Women on the Cusp of the 21st Century, 26 Ohio N.U. L. Rev. 581, 584 (2000) (finding high rates of commutations and judicial intervention in capital cases with female defendants). The findings on the basis of race, however, are more complicated. In the period before the Supreme Court’s 1972 decision in Furman v. Georgia, 408 U.S. 238 (1972) (per curiam), researchers generally found that white people on death row were more likely to receive clemency. See, e.g., James W. Marquart, Sheldon Ekland-Olson & Jonathan R. Sorensen, The Rope, the Chair, and the Needle: Capital Punishment in Texas, 1923–1990, at 117 tbl.5.1 (1994) (finding that Anglo and Hispanic death row prisoners benefitted from clemency more than African-Americans). In the period after Furman, studies have not found that white people are more likely to receive a commutation in capital cases and they may, in fact, be less likely to receive one. John Kraemer, An Empirical Examination of the Factors Associated with the Commutation of State Death Row Prisoners’ Sentences Between 1986 and 2005, 45 Am. Crim. L. Rev. 1389, 1410 (2008) (“Prisoners of black, Hispanic, or other racial/ethnic heritage have slightly over twice the odds of commutation . . . compared to their white counterparts . . . .” (footnote omitted)); Michael Heise, The Death of Death Row Clemency and the Evolving Politics of Unequal Grace, 66 Ala. L. Rev. 949, 974 (2015) (“The probability of African-American death row inmates receiving clemency significantly exceeds that of non-African-Americans.”). One possible reason for this difference is that racial disparities at earlier stages might mean that the average African-American person on death row committed a relatively less serious offense than the average white person on death row and is therefore a better candidate for clemency. Id.
  18. . Gunderson, supra note 17, at 2.
  19. . Factbox: Balance of Mississippi Pardons by Race, Reuters (Jan. 20, 2012, 5:43 PM), https://www.reuters.com/article/idUSTRE80J25R/ [https://perma.cc/WTJ6-HWCK].
  20. . Dafna Linzer & Jennifer LaFleur, Presidential Pardons Heavily Favor Whites, ProPublica (Dec. 3, 2011, 11:00 PM), https://www.propublica.org/article/shades-of-mercy-presidential-forgiveness-heavily-favors-whites [https://perma.cc/9PJL-6Z3P].
  21. . Mark Osler, The Trump Clemencies: Celebrities, Chaos, and Lost Opportunity, 31 Wm. & Mary Bill Rts. J. 487, 515 (2022).
  22. . See, e.g., Jeffery T. Ulmer, Michael T. Light & John H. Kramer, Racial Disparity in the Wake of the Booker/FanFan Decision: An Alternative Analysis to the USSC’s 2010 Report, 10 Crimonology & Pub. Pol’y 1077, 1077, 1080 (2011) (noting bias toward white defendants in the in-out decision); Tina L. Freiburger & Carly M. Hilinski, An Examination of the Interactions of Race and Gender on Sentencing Decisions Using a Trichotomous Dependent Variable, 59 Crime & Delinq. 59, 63–64 (2013) (finding that white defendants in Philadelphia were more likely to get probation instead of incarceration); Ryan D. King & Brian D. Johnson, A Punishing Look: Skin Tone and Afrocentric Features in the Halls of Justice, 122 Am. J. Socio. 90, 97, 115 (2016) (concluding that defendants with darker skin tones were treated more punitively in court in Ramsey and Hennepin Counties in Minnesota).
  23. . See, e.g., M. Marit Rehavi & Sonja B. Starr, Racial Disparity in Federal Criminal Sentences, 122 J. Pol. Econ. 1320, 1349 (2014) (controlling for arrest offense and criminal history and finding that Black defendants received sentences about 9% longer than similarly situated white defendants); Ulmer et al., supra note 22, at 1100–01 (finding that both male and female Black defendants were less likely to receive a downward sentencing departure than white men).
  24. . Darrell Steffensmeier & Stephen Demuth, Does Gender Modify the Effects of Race–Ethnicity on Criminal Sanctioning? Sentences for Male and Female White, Black, and Hispanic Defendants, 22 J. Quantitative Criminology 241, 242, 256 (2006) (analyzing sentencing data from fifty-four urban counties from 1990–1996).
  25. . Michael L. Radelet & Glenn L. Pierce, Race and Death Sentencing in North Carolina, 19802007, 89 N.C. L. Rev. 2119, 2137, 2140 (2011).
  26. . John J. Donohue III, Capital Punishment in Connecticut, 1973-2007: A Comprehensive Evaluation from 4,686 Murders to One Execution, in Selected Works of John Donahue 1 (2011), https://dpic-cdn.org/production/legacy/DonohueCTStudy.pdf [https://perma.cc/VN6C-8UNQ].
  27. .Id. at 7.
  28. . Jennifer L. Eberhardt, Paul G. Davies, Valerie J. Purdie-Vaughns & Sheri Lynn Johnson, Looking Deathworthy: Perceived Stereotypicality of Black Defendants Predicts Capital-Sentencing Outcomes, 17 Psych. Sci. 383, 384 (2006).
  29. . Alison Siegler, End Mandatory Minimums, Brennan Ctr. for Just. (Oct. 18, 2021), https://www.brennancenter.org/our-work/analysis-opinion/end-mandatory-minimums [https://perma.cc/HS8A-PNB8] (noting that “[b]y the end of the 1980s, all 50 states had enacted mandatory minimums” and many were enacted at the federal level, too); Rachel E. Barkow & Kathleen M. O’Neill, Delegating Punitive Power: The Political Economy of Sentencing Commission and Guideline Formation, 84 Texas L. Rev. 1973, 1994 tbl.1 (2006) (listing years in which sentencing commissions and guidelines were instituted).
  30. . Marc Mauer, Why Are Tough on Crime Policies so Popular?, 11 Stan. L. & Pol’y Rev. 9, 11 (1999).
  31. . Rachel E. Barkow, Administering Crime, 52 UCLA L. Rev. 715, 742 (2005); Edward M. Kennedy, Toward a New System of Criminal Sentencing: Law with Order, 16 Am. Crim. L. Rev. 353, 356–57 (1979); Donald W. Dowd, What Frankel Hath Wrought, 40 Vill. L. Rev. 301, 302–303 (1995).
  32. . Stephen Breyer, The Federal Sentencing Guidelines and the Key Compromises upon Which They Rest, 17 Hofstra L. Rev. 1, 4–5 (1988); Ken Feinberg, Federal Criminal Sentencing Reform: Congress and the United States Sentencing Commission, 28 Wake Forest L. Rev. 291, 295–96 (1993); Marvin E. Frankel, Criminal Sentences: Law Without Order 76 (1973).
  33. . Chad M. Topaz, Shaoyang Ning, Maria-Veronica Ciocanel & Shawn Bushway, Federal Criminal Sentencing: Race-Based Disparate Impact and Differential Treatment in Judicial Districts, 10 Hums. & Soc. Scis. Commc’ns, no. 366, 2023, at 1, 2; Tushar Kansal, The Sent’g Project, Racial Disparity in Sentencing 9 (Marc Mauer ed., 2005); Cassia C. Spohn, Thirty Years of Sentencing Reform: The Quest for a Racially Neutral Sentencing Process, in 3 Policies, Processes, and Decisions of the Criminal Justice System 427, 428 (2000).
  34. . See Rehavi & Starr, supra note 23, at 1323 (finding that the odds of Black defendants being charged with a crime carrying a mandatory minimum was 1.75 times higher than the odds of white defendants being so charged). A study of the three-strikes law in California found similar biases in prosecutorial charging discretion, with prosecutors being more likely to charge so-called wobbler offenses as felonies when the defendant is Black and as misdemeanors when the defendant is white, thus triggering the three-strikes law against Black defendants in more cases. Elsa Y. Chen, The Liberation Hypothesis and Racial and Ethnic Disparities in the Application of California’s Three Strikes Law, 6 J. Ethnicity in Crim. Just. 83, 92, 94 (2008).
  35. . U.S. Sent’g Comm’n, Fifteen Years of Guidelines Sentencing 135 (2004).
  36. . 50-State Comparison: Pardon Policy & Practice, Restoration of Rts. Project (Oct. 2023), https://ccresourcecenter.org/state-restoration-profiles/50-state-comparisoncharacteristics-of-pardon-authorities-2/ [https://perma.cc/NVL5-2SJP].
  37. . See Pa. Const. art. 4, § 9 (adding additional requirements for pardons via an amendment in 1997).
  38. . Ctr. on the Admin. of Crim. L., N.Y.U., The Demise of Clemency for Lifers in Pennsylvania 4, 8 (2023), https://www.law.nyu.edu/sites/default/files/CACL%20Clemency
    %20PA_Accessible.pdf [https://perma.cc/TS4F-GMY2].
  39. . Id. at 4.
  40. . Kelan Lyons, Connecticut’s New Commutation Policy Raises the Bar for Second Chances, Bolts Mag. (Sept. 8, 2023), https://boltsmag.org/connecticut-commutation-policy/ [https://perma.cc/T7N4-G3JW].
  41. . Id.
  42. . Jamiles Lartey, Connecticut Normalized Clemency. Not Anymore., The Marshall Project (May 6, 2023, 12:00 PM), https://www.themarshallproject.org/2023/05/06/connecticut-incarceration-clemency-commutation-pardon-justice-reform [https://perma.cc/TB2E-CFZ7].
  43. . State of Conn. Bd. of Pardons & Paroles, Policy No. III.02, Commutations 2, 4 (2023), https://portal.ct.gov/-/media/BOPP/Pardons/Policy-III02-Commutations.pdf [https://perma.cc/V5RG-8BYC].
  44. . Because there are so many structural biases that produce differential offending rates on the basis of race, solving the problem of unwarranted racial disparities in the dispensation of grants of mercy will not solve racial disparities overall. The goal here is to make sure that grants of mercy do not exacerbate the disparities that already exist and perhaps can help mitigate them.
  45. . Nicolas M. Pace, James M. Anderson, Shamena Anwar, Danielle Schlang, Melissa A. Bradley & Amalavoyal Chari, Statistical Analysis of Presidential Pardons 42–69 (Bureau of Just. Stat., Working Paper No. NCJ 300116, 2021), https://www.ojp.gov/pdffiles1/bjs/grants/300116.pdf [https://perma.cc/94CR-YL6Z].
  46. . Id.; see also Margaret Love, Study Reveals Potential for Racial Bias in Presidential Pardon Process, Collateral Consequences Res. Ctr. (June 24, 2021), https://ccresourcecenter.org
    /2021/06/24/study-reveals-potential-for-racial-bias-in-presidential-pardon-process/ [https://perma.cc/P5WQ-3JEC] (explaining that white applicants are more likely to satisfy the Office of the Pardon Attorney’s standards for recommending a grant to the president).
  47. . Off. of the Pardon Att’y, Obama Administration Clemency Initiative, Dep’t of Just., https://www.justice.gov/archives/pardon/obama-administration-clemency-initiative [https://perma.cc/8NGV-P8V2].
  48. . U.S. Sent’g Comm’n, An Analysis of the Implementation of the 2014 Clemency Initiative 14 (2017), https://www.ussc.gov/sites/default/files/pdf/research-and-publications
    /research-publications/2017/20170901_clemency.pdf [https://perma.cc/YR5A-VJ2V] (noting that 61% of commutation grantees had been sentenced for a crack cocaine trafficking offense). Although the explicit announcement did not specify this, the actual focus of the initiative was people serving sentences for drug crimes, and all the people who ultimately received clemency were incarcerated for drug trafficking offenses. Rachel E. Barkow & Mark Osler, Designed to Fail: The President’s Deference to the Department of Justice in Advancing Criminal Justice Reform, 59 Wm. & Mary L. Rev. 387, 429–30, 430 n.190 (2017); Bureau of Just. Stat., Drug Offenders in Federal Prison: Estimates of Characteristics Based on Linked Data Summary 1 (2015), https://bjs.ojp.gov/content/pub/pdf/dofp12_sum.pdf [https://perma.cc/3QV5-257D].
  49. . If the goal is to correct excessive sentences, this factor is, however, too narrow. It leaves out the people who would still receive an excessively long sentence today because of overzealous prosecution. Barkow & Osler, supra note 48, at 430.
  50. . Rachel Elise Barkow, Prisoners of Politics: Breaking the Cycle of Mass Incarceration 44–47 (2019) (citing research on the diminishing returns of long sentences and their sometimes-criminogenic effects).
  51. . Off. of the Pardon Att’y, supra note 47.
  52. . Barkow & Osler, supra note 48, at 430.
  53. . U.S. Sent’g Comm’n, supra note 48, at 13.
  54. . U.S. Sent’g Guidelines Manual § 1B1.13 (U.S. Sent’g Comm’n 2023), https://www.ussc.gov/sites/default/files/pdf/guidelines-manual/2023/GLMFull.pdf [https://perma.cc/2PTR-4YJM].
  55. . While the categorical eligibility requirements have helped address racial disparities, compassion release overall suffers from wide disparity by geographic district, with federal judges differing widely in their views of how broadly to use their authority. For example, the Sentencing Commission studied compassionate release decisions during fiscal year 2020 and found “considerable variability,” ranging from judges in the First Circuit granting 47.5% of compassionate release requests to judges in the Fifth Circuit granting only 13.7% of requests. U.S. Sent’g Comm’n, Compassionate Release: The Impact of the First Step Act and COVID-19 Pandemic 4 (2022), https://www.ussc.gov/sites/default/files/pdf/research-and-publications
    /research-publications/2022/20220310_compassionate-release.pdf

    [https://perma.cc/UL94-WN4S].

  56. . Id. at 22.
  57. . McCleskey v. Kemp, 481 U.S. 279, 355 (1987) (Blackmun, J., dissenting).
  58. . Id. at 325 n.2 (Blackmun, J., dissenting).
  59. . See Dan Bang & Chris D. Frith, Making Better Decisions in Groups, Royal Soc’y Open Sci., Aug. 2017, at 1, 7–8 (discussing the problems that arise when group members have similar past experiences, backgrounds, and personal characteristics).
  60. . See Rachel E. Barkow & Mark Osler, Restructuring Clemency: The Cost of Ignoring Clemency and a Plan for Renewal, 82 U. Chi. L. Rev. 1, 21–22 (2015) (discussing the benefits of a diverse board for clemency).
  61. . For a discussion of how this can be done in the context of sentencing commissions faced with political pressures, see Barkow, supra note 50, at 169–175.