Data, Disparities, and Discrimination: How California’s Racial Justice Act Creates New Pathways to Challenge and Evaluate Racial Bias

California state capitol museum

ANALYSIS BRIEF

Data, Disparities, and Discrimination

How California’s Racial Justice Act Creates New Pathways to Challenge and Evaluate Racial Bias

Dan Sutton

Stanford Center for Racial Justice
Published: April 15, 2025

Download PDF

KEY TAKEAWAYS

California’s Racial Justice Act directly responds to McCleskey v. Kemp by explicitly allowing statistical evidence of racial disparities in criminal proceedings. This represents California’s attempt to reopen courthouse doors that the U.S. Supreme Court effectively closed in 1987, creating new pathways for defendants to challenge racial bias without proving intentional discrimination in their cases.

Using statistical evidence poses implementation challenges when analyzing whether defendants were treated differently than “similarly situated” individuals of other races. Key challenges include creating appropriate comparison groups, handling small sample sizes, and determining what constitutes a significant racial disparity.

Data availability impacts implementation as effective analysis depends on robust prosecutorial data that has historically been limited. The Act’s requirement that prosecutors provide underlying data may incentivize better collection practices, potentially creating a model that other states may adopt to address racial bias in their criminal justice systems.

California’s criminal courtrooms are seeing a new kind of evidence. Where data analysis was once limited to specialized areas like forensic DNA probabilities or challenges to discriminatory jury selection, statistical evidence is now poised to play a central role in many criminal cases thanks to a relatively new law: the California Racial Justice Act.1 Enacted in 2020, the law allows defendants to challenge their criminal charges or sentences using evidence of racial bias or discrimination, including statistical evidence.2 Amendments in 2022 and 2023 have expanded the law’s reach and given it retroactive application.3

This brief examines how courts, prosecutors, and defense attorneys are interpreting and applying the Act’s provisions, with particular attention to the methodological challenges of using statistical evidence of racial disparities. Our aim is to provide a framework for understanding these emerging issues and to offer guidance to practitioners navigating this new legal landscape.

The Racial Justice Act represents California’s direct response to the U.S. Supreme Court’s 1987 decision in McCleskey v. Kemp, which effectively closed the courthouse doors to statistical evidence of racial disparities in criminal proceedings.4 By explicitly allowing defendants to use data to demonstrate bias, the Act creates a considerable shift in how discrimination can be challenged in the criminal justice system.

The significance of this shift is evident in how quickly the state’s criminal justice institutions have engaged with the Act’s provisions. In Los Angeles, the nation’s largest public defender’s office has created a new unit focused on bringing claims under the law.5 Our review of public records shows that prosecutors’ offices in at least 15 of California’s 58 counties have contracted with the same consulting firm to analyze data related to RJA claims, at a cost of more than $6 million to taxpayers.6 The University of San Francisco Law School has received a $1 million donation to expand its Racial Justice Clinic to support litigation under the Act.7

The energy around the law stems partly from its straightforward premise: show evidence of racial bias or discrimination and the Act directs courts to do things like declare mistrials, reduce charges, or dismiss sentence enhancements. Some provisions, like sections (a)(1) and (a)(2), address the use of racially discriminatory language–by a judge, an attorney, the police, a witness, or even a juror–and, on its face, identifying such evidence appears uncomplicated.9 Take, for instance, a police officer’s text to a colleague: “I’m only stopping them cuz they black.”10

Other provisions of the law require defense attorneys, prosecutors, and judges–who typically lack training in statistical analysis–to craft, contest, and evaluate complex data-driven arguments.11 These sections, specifically (a)(3) and (a)(4) allow defendants to use charging and sentencing data to show that they were treated differently than other “similarly situated” individuals of other races. The statute explicitly states that “the defendant does not need to prove intentional discrimination,” though this terminology raises complex questions about how discrimination operates in practice.12 Data from prosecutors’ offices, courts, and police departments appears sufficient on its own to establish a violation of the Act. But, like many statutes, the law provides little guidance on how this analysis should be conducted, leaving fundamental questions unanswered– including how to determine which defendants are truly “similarly situated.”

Both the data-driven claims, on which this brief focuses, and the language-based claims highlight common conceptual questions, particularly regarding the meaning of “intentional discrimination” and the relevance of causation, that help reveal how courts are interpreting the Act’s provisions. Our analysis centers on statistical evidence as it presents the most significant departure from existing law and a set of unique interpretive challenges for courts and practitioners.

Even with these unanswered questions and the Act’s relative newness, the RJA is already impacting cases moving through California’s criminal justice system. In December 2024, an appeals court overturned the rape conviction of former San Francisco 49ers defensive lineman Dana Stubblefield after finding that the prosecutor’s statements during closing arguments appealed to racial bias and violated the Act.13 The court determined that the prosecutor’s suggestion that police avoided searching Stubblefield’s home for a gun following George Floyd’s murder improperly invited jurors to consider the defendant’s race when weighing evidence, potentially creating a perception that Stubblefield received an advantage due to being Black.14 In another case, statistical evidence showing the overrepresentation of Black defendants in Contra Costa County gang-murder charges led to the dismissal of enhanced “special circumstance” murder charges against four Black defendants.15 In that same case, where Antioch, California police officers exchanged a series of racist text messages, the court later found that eight officers had exhibited racial bias under the Act and dismissed all gang-related special allegations against the defendants.16

The Stanford Center for Racial Justice is committed to research and analysis that advances the understanding of emerging issues at the intersection of race, bias, and America’s justice system. Along with our newly developed Toolkit on Statistical Claims Under the California Racial Justice Act, created with colleagues from the Justice Innovation Lab, this brief aims to contribute to a fundamental aim behind the RJA: identifying and removing racial bias in the criminal justice system where it exists–an objective that should transcend political lines and courtroom roles.

UNDERSTANDING THE RACIAL JUSTICE ACT: OPENING COURTHOUSE DOORS TO DATA

To better understand the California Racial Justice Act we return to McCleskey v. Kemp–the Supreme Court case challenging the state of Georgia’s use of the death penalty on racial discrimination grounds.17 A previously published ‘Explainer’ from the Stanford Center for Racial Justice detailed the case’s background: Warren McCleskey, a Black man convicted of robbing an Atlanta furniture store and murdering a white police officer, presented an empirical study showing dramatic disparities in Georgia’s death sentences.18

This study, known as the Baldus study, was extraordinarily comprehensive, accounting for more than 200 non-racial variables that might have explained the disparities in sentencing outcomes.19 The data revealed that defendants charged with killing white victims were 4.3 times more likely to be sentenced to death, and, even more striking, Black defendants with white victims were sentenced to death at nearly 22 times the rate of Black defendants with Black victims.20 The Court’s rejection of such a methodologically rigorous statistical analysis highlights the substantial barrier it created for addressing racial disparities in the criminal justice system.

Even while acknowledging the study indicated “a discrepancy that appears to correlate with race,” Justice Lewis Powell, writing for the majority, dismissed the evidence of disparities in sentencing as “an inevitable part of our criminal justice system.”21 Unlike earlier rulings that credited statistical proof, McCleskey set a higher bar for criminal defendants, holding that even strong statistical evidence would not prove a constitutional violation.22 The Court also tightened its standard for showing discrimination laid out in earlier cases–a disparate racial impact and a discriminatory purpose–by insisting on evidence of intentional discrimination in the specific case.23 The impact has been profound and lasting. As legal scholar Michelle Alexander observes, the ruling has made it “virtually impossible” to challenge racial bias in the criminal justice process without evidence of intentional discrimination–evidence that is difficult to come by.24 “The U.S. Supreme Court,” Alexander explains, “has said that the courthouse doors are closed to claims of racial bias.”25 California’s Racial Justice Act aims to change that.

Two Pathways to Challenge Bias: How Language-based and Data-driven Claims Raise Common Questions

The Act creates several pathways for defendants to challenge their criminal charges or sentences using evidence of racial bias or discrimination. Defendants can point to language used by people involved in their case, expert testimony, and prosecutorial data on “similarly situated” individuals to advance RJA claims.26 Notably, the Act also requires prosecutors to provide defendants with the data and evidence needed to support their claims upon request.27

Under sections (a)(1) and (a)(2), a defendant can present evidence that a judge, attorney, police officer, expert witness, or juror involved in their case exhibited racial bias or used “racially discriminatory language” that explicitly or implicitly appeals to racial bias.28 This seems straightforward. Lawyers and judges should, in theory, be able to identify racist language when they see it, and the Act directs courts to take corrective action such as dismissing charges or vacating sentences. Yet without an agreed-upon baseline for what constitutes racial bias in the view of an “objective observer”–the standard the Act establishes– courts have reached varying conclusions. They have ruled that prosecutors’ references to non-white defendants as “monsters” and “predators” did not violate the Act, while finding racial bias in a prosecutor’s assertion that police deliberately avoided searching a famous Black defendant’s house in the weeks following George Floyd’s murder.29 In another case, People v. Johnson, courts deemed permissible an expert’s repeated use of “gorilla pimp” during a sex trafficking trial of two Black defendants, accepting the argument that the term aimed to convey victims’ perspectives.30

Data-driven claims under sections (a)(3) and (a)(4) require more complex analysis. These provisions permit defendants to show they were charged, convicted, or sentenced more severely than defendants of other races in the county.31 Comparisons must be made between “similarly situated” individuals who have engaged in similar conduct. While the law calls for “statistical evidence” or “aggregate data” demonstrating a “significant difference” in how people of different races are treated, it offers little guidance on how to construct these comparisons.32

The Racial Justice Act states that defendants do “not need to prove intentional discrimination” when bringing claims.33 This language aims to lower the barrier McCleskey erected. But what does “intentional discrimination” actually mean in this context?

On its face, this term might refer to evidence that a specific decision-maker in a defendant’s case like a prosecutor or juror acted on the basis of deliberate racial bias. But this narrow interpretation may not fully account for how discrimination actually functions in the criminal justice system.

Modern conceptions of discrimination recognize that bias operates on multiple levels. There’s explicit bias, where someone consciously acts on racial prejudice. But there’s also implicit or unconscious bias, where decision-makers may be influenced by racial factors without their conscious awareness. Beyond individual bias, there are also systemic patterns and institutional practices that might produce racially disparate outcomes regardless of any individual’s intent.

When statistical analysis under sections (a)(3) and (a)(4) controls for all relevant non-racial variables and still finds differences by race, the implication is that the treatment of race itself is influencing outcomes. This suggests that some form of discrimination is active in the process, whether through explicit policies, unconscious biases, or embedded systemic practices. In other words, when rigorous analysis reveals persistent racial disparities, the statistical evidence itself indicates that a prosecutor’s decisions to charge particular crimes or seek specific sentences are being influenced by race-based discrimination, rather than these being separate considerations.

Similarly, the language-based provisions in sections (a)(1) and (a)(2) focus on whether a person “exhibited bias or animus” or used “racially discriminatory language,” behaviors that either reflect conscious prejudice or unconscious bias, regardless of whether the speaker intended to discriminate in the legal sense that McCleskey required.34

Courts’ analysis of language-based RJA claims reveals an underlying tension about causation—whether bias must directly impact case outcomes to violate the Act, or whether the Act aims to eliminate racist language regardless of proven harm. Despite the statute’s focus on the mere presence of discriminatory language, courts often find themselves analyzing potential causal connections. In the Stubblefield case where the prosecutor referenced George Floyd in the closing argument, while the court noted “the RJA does not require showing how an appeal to racial bias might have impacted a defendant’s trial,” it immediately went on to observe how the prosecutor’s statements could have influenced jurors who might feel compelled to ‘push back’ on anti-police sentiment when considering a Black defendant’s case.35

In the Johnson case–where the expert used the term “gorilla pimp” during the sex trafficking trial of two black defendants, the court permitted this language partly because it appeared “strictly in the context of [the expert] providing background information” rather than characterizing the defendants directly.36 The court also emphasized that “the most damning evidence in the case against defendants was their own words,” suggesting that even if the language were problematic, it likely didn’t affect the outcome.37 These rulings illustrate how difficult it is for courts to separate the mere presence of racially coded language from questions about its impact—a difficulty that reflects the deeper ambiguity about whether the RJA aims to purge all racially charged language from criminal proceedings or only language that clearly disadvantages defendants.

Few courts have yet evaluated statistical claims under the Racial Justice Act. In the Antioch, California case–People v. Windom–a trial court judge found that data from approximately 90 gang-murder cases between 2015-2022 demonstrated that the Contra County District Attorney’s Office disproportionately applied “special circumstances” charging enhancements against Black defendants.38 When the prosecution failed to provide a race-neutral explanation for this disparity, as required by the Act’s burden-shifting framework, the court dismissed the enhanced charge for the four defendants.39 Similarly, an appellate court reviewing Riverside County’s use of the death penalty against Black defendants in People v. Mosby determined that a combination of facts from similar cases and statistical evidence was sufficient to establish an initial, prima facie case, but declined to decide whether statistical evidence alone could meet this threshold.40

The Racial Justice Act appears designed to serve two important functions in California’s criminal justice system. First, it lowers the bar for defendants to challenge racial disparities in the prosecutorial process. Second, and perhaps equally significant, the law may function as a “best practices” nudge, incentivizing prosecutors’ offices to develop more systematic, transparent procedures for charging decisions.41 Much like how employment discrimination law has encouraged employers to implement clearer job descriptions and reform hiring practices to minimize legal exposure, the RJA could prompt prosecutors to establish more consistent charging guidelines and documentation processes. These early decisions mark just the beginning of California courts’ engagement with statistical evidence under the Act, but they may already be driving procedural improvements in the system.

EMERGING ISSUES IN DEVELOPING AND EVALUATING STATISTICAL CLAIMS UNDER THE RACIAL JUSTICE ACT

Data-driven claims under the Racial Justice Act are expected to multiply in California’s courts. The law’s structure, combined with current policing and prosecutorial practices, seems likely to place several key questions at the center of this litigation: how to create appropriate comparison groups, how to handle small sample sizes, and how to determine whether any observed racial or ethnic disparities are significant. Where Racial Justice Act precedents remain limited, criminal practitioners can draw valuable insights from the extensive body of case law on data as evidence in civil rights litigation, particularly in employment and housing discrimination cases.

Federal courts have articulated a basic principle for assessing disparate treatment: analysis must “compare the group that ‘enters’ the process with the group that emerges from it.”42 This means looking beyond general population statistics to identify the particular group affected by a specific policy or decision— an approach the RJA adopts through its focus on “similarly situated” individuals who engage in similar conduct.43 But choosing this comparison group can require careful consideration of decision points that may not be immediately apparent.44 For instance, in examining racial disparities in gang enhancements, meaningful analysis might consider not just cases where gang charges were filed–as was done in the Windom decision–but all cases where these charges could have been filed based on evidence of gang activity and the underlying felony offenses. Just as employment discrimination analysis often examines the entire pool of promotion-eligible employees rather than just those promoted, defendants may argue that reviewing all cases eligible for prosecution provides a more complete picture of prosecutorial decision-making.45

Law enforcement policies or practices can also skew or limit the availability of straightforward data for comparisons. Consider a police department’s decision to more aggressively police particular neighborhoods. More cases from those neighborhoods will wind up on a prosecutor’s desk and simply looking at how prosecutors handle these cases may mask underlying disparities in how cases enter the system.46 This dynamic parallels housing discrimination cases where a landlord’s reputation for discrimination may deter applications altogether, making official application data an incomplete measure of discrimination’s true impact. The Racial Justice Act acknowledges this complexity, directing courts to consider whether, “systemic and institutional racial bias, racial profiling, and historical patterns of racially biased policing and prosecution” may have contributed to observed disparities or impacted data availability.47 For this reason, attorneys litigating RJA claims are likely to bring in other, broader data sources–such as comparing arrest rates to underlying crime data across neighborhoods, or comparing similar jurisdictions with different policing practices–to see whether certain groups are being disproportionately pushed into the criminal justice system.

When groups of similarly situated defendants are small– whether due to a jurisdiction’s size or how rarely certain charges are filed–they don’t automatically preclude findings of discrimination but require careful analysis.48 Federal courts analyzing employment discrimination claims have contended with this challenge in cases involving small offices or departments.49 These courts have often looked at disparities based on small samples with skepticism, particularly when minor changes would dramatically change the statistical picture, or even reverse the disparity.50 Sample size issues can be further complicated by the periods used for analysis. Racial Justice Act claims must necessarily focus on specific time periods, and the shorter that window is, the smaller the universe of relevant cases. And, over time, leadership changes in district attorney offices and police departments can shift policies and practices–changes that are often poorly documented but may impact data patterns.

Evaluating statistical significance in Racial Justice Act claims is also likely to generate considerable litigation, particularly around what constitutes a meaningful racial disparity. While no rigid mathematical threshold is required, judges typically evaluate the substantiality of disparities on a case-by-case basis.51 Many federal courts regularly reference analytical approaches from employment discrimination law like the “four-fifths rule,” which identifies potential discrimination when one group’s ‘selection rate’ falls below 80% of another group’s rate.52 For example, if prosecutors in a county did not seek the death penalty in 90% of White defendants’ cases but only in 70% of Black defendants’ cases, the Black selection rate would be 78% of the White rate, below the 80% threshold, suggesting a potentially significant disparity.

However, scholars have observed that courts, including the Supreme Court, have been notably inconsistent in how they treat statistical evidence across different contexts.53 Chief Justice William Rehnquist, who joined the majority’s rejection of the Baldus study in McCleskey, sixteen years later in Grutter v. Bollinger relied heavily on statistical evidence in his dissent to argue that the University of Michigan Law School had discriminated against White applicants.54 In the Racial Justice Act, California courts may be prompted to chart a different course–one that moves beyond the federal system’s apparent civil-criminal divide and establishes a consistent framework for evaluating statistical evidence of discrimination across all areas of the law.

THE ROLE OF PROSECUTORIAL DATA: SHAPING THE FUTURE OF RACIAL JUSTICE ACT CLAIMS

One particular challenge hovers over nearly every aspect of data-driven claims under sections (a)(3) and (a)(4) of the Racial Justice Act–the provisions allowing defendants to use statistical evidence–the availability and quality of prosecutorial data. Effective analysis of these claims, whether formulating, countering, or evaluating them, depends on robust, accurate data, yet prosecutors have historically been slow to collect, organize, and analyze information about important decision points in the processing of a case.55

This landscape is shifting. A 2018 national survey of prosecutors’ offices by Urban Institute researchers revealed that most offices are collecting some key data measures, with approximately 40 percent collecting all seven measures identified by the researchers as foundational to tracking prosecutorial activities like charging decisions, case declinations, and dismissals.56 But these high-level data points barely scratch the surface of how the prosecutorial process operates.

Take the plea bargaining process, where at least 97% of all California state criminal convictions are obtained.57 Despite its pivotal role in the criminal justice system, little to no data about the bargaining process is formally recorded, with information about prior plea offers often existing only as handwritten notes scattered throughout prosecutors’ case files.58 In 2023, the American Bar Association’s Plea Bargain Task Force called for a significant change: recording the history of plea offers in each individual case in searchable databases accessible to prosecutors, defense counsel, and judges–an approach far removed from current practice but potentially valuable in implementing the Racial Justice Act’s aims.59

California courts may look unfavorably at situations where prosecutors’ offices cannot produce data relevant to (a)(3) or (a)(4) claims. There is some precedent for such skepticism: In Lehman v. Trout, an employment discrimination case against the U.S. Navy that ultimately reached the Supreme Court, both trial and appellate courts noted that plaintiffs “cannot legitimately be faulted for gaps in their statistical analysis when the information necessary to close those gaps was possessed only by defendants and was not furnished either to plaintiffs or to the Court.”60 As RJA litigation evolves, prosecutors’ offices that develop more comprehensive data collection practices and facilitate easier access to this information may not only better respond to claims unsupported by evidence but also contribute to more meaningful evaluations of racial disparities where they exist.

CONCLUSION

As California’s courts grapple with implementing the Racial Justice Act’s data-driven provisions, the challenges extend beyond the legal and analytical frameworks discussed in this brief. Practical hurdles loom: statistical evidence doesn’t analyze itself and the effective working of the Act depends on access to skilled experts capable of sophisticated analysis and credible testimony about their findings. The scale of this challenge is already apparent–our review of public records reveals that prosecutors’ offices in at least 15 of California’s 58 counties have entered into millions of dollars of contracts with a single consulting firm–Sicuro Data Analytics–to analyze RJA-related data.61 Substantial costs could stretch an already resource-strained criminal justice system, particularly public defender offices. And, while the Act also allows courts to appoint their own independent experts, a solution some scholars have advocated for in other contexts, it’s unclear whether this approach is sustainable across a wide number of cases.62

These practical constraints, however, should not overshadow the Act’s far-reaching potential. By explicitly allowing statistical evidence to prove racial bias—and requiring prosecutors to provide the underlying data, in turn potentially incentivizing better data collection—California has created a framework that could significantly alter how criminal courts evaluate discrimination claims. The early cases discussed in this brief suggest that, despite the challenges, courts and practitioners are taking the first tentative steps toward addressing how statistical evidence of racial disparities might be analyzed and presented, though much remains to be determined. As this body of law continues to evolve, it may offer a model for other jurisdictions seeking to move beyond McCleskey’s limitations and create more robust mechanisms for evaluating and addressing racial bias in criminal justice.

 

 

REFERENCES

  1. See Christopher Slobogin, “The use of statistics in criminal cases: An introduction,” Behavioral Sciences & the Law (Mar. 27, 2019); Graham R. Cronogue, “Lies, Damn Lies, and Batson Challenges: The Right to Use Statistical Evidence to Prove Racial Bias,” University of Miami Race & Social Justice Law Review (Aug. 1, 2016); National Research Council Committee on DNA Forensic Science: An Update. “The Evaluation of Forensic DNA Evidence,” National Academies Press (1996).
  2. CAL. PENAL CODE § 745 (West 2025).
  3. Id. (enacted by 2020 Cal. Stat. ch. 317 (A.B. 2542), amended by 2022 Cal. Stat. ch. 739 (A.B. 256), 2023 Cal. Stat. ch. 464 (A.B. 1118), and 2024 Cal. Stat. ch. 495 (S.B. 1518)).
  4. McCleskey v. Kemp, 481 U.S. 279, 325 (1987).
  5. Los Angeles County Public Defender’s Office, “Racial Justice,” (accessed Mar. 5, 2025).
  6. Author’s calculations based on review of public records of agreements between various California prosecutors’ offices and Sicuro Data Analytics. See Emi MacLean, “Embracing “Too Much Justice”: Realizing the Potential of the California Racial Justice Act,” Berkeley Journal of Criminal Law (2024) for discussion of Sicuro’s role in responding to a Racial Justice Act-related data request to Tulare County.
  7. Sara Rinaldi, “A Gift of $1 Million to USF School of Law’s Racial Justice Clinic,” (May 31, 2024).
  8. Andrew Cohen, “Implementing Equality: Packed Symposium Addresses California Racial Justice Act,” UC Berkeley Law (Feb. 13, 2024) describing how the “massive turnout” for a Racial Justice Act symposium “signaled the topic’s importance to lawyers across the state…”
  9. PENAL § 745(a)(1) and (a)(2).
  10. Larry J. Wallace, “Investigation Report,” Office of the District Attorney Contra Costa County 20 (accessed Mar. 27, 2023).
  11. PENAL § 745(a)(3) and (a)(4).
  12. PENAL § 745(b)(2).
  13. People v. Stubblefield, H048598 (Cal. Ct. App. 2024).
  14. Id.
  15. Court’s Order Re: 745(a)(3) Motion at 4, People v. Windom et al., No. 01001976380, (Cal. Ct. Contra Costa Cnty. fled May 23, 2023).
  16. See Court’s Order Re: 745(a)(1) Motion at 1-2, People v. Windom et al., No 01001976380, (Cal. Ct. Contra Costa Cnty. fled Feb. 5, 2024).
  17. See Ash Kalra, “California Racial Justice Act for All Signed Into Law,” (Sep. 30, 2022).
  18. Hoang Pham and Amira Dehmani, “The California Racial Justice Act of 2020, Explained,” Stanford Center for Racial Justice (Apr. 22, 2024).
  19. David C. Baldus, Charles Pulaski, and George Woodworth, “Comparative Review of Death Sentences: An Empirical Study of the Georgia Experience,” 74 J. Crim. L. & Criminology 661 (1983).
  20. McCleskey v. Kemp, 481 U.S. 279, 325 (1987) (Brennan, J., dissenting).
  21. McCleskey v. Kemp, 481 U.S. 279 (1987).
  22. See Reva B. Siegel, Opening Address “Blind Justice: Why the Court Refused to Accept Statistical Evidence of Discriminatory Purpose in McCleskey v. Kemp–And Some Pathways for Change,” Northwestern University Law Review’s 2017 Symposium: “A Fear of Too Much Justice”? Equal Protection and the Social Sciences 30 Years After McCleskey v. Kemp (2016).
  23. See Mario L. Barnes and Erwin Chemerinsky, “What Can Brown Do for You?: Addressing McCleskey v. Kemp as a Flawed Standard for Measuring the Constitutionally Significant Risk of Race Bias,” Northwestern Law Review (2018).
  24. Michelle Alexander, appearing with Bryan Stevenson and Bill Moyers (April 2, 2010). “Bill Moyers Journal.”
  25. Id.
  26. PENAL § 745(a).
  27. Id. § 745(d).
  28. Id. § 745(a)(1) and (a)(2).
  29. People v. Quintero, A165276 (Sonoma Cnty. Super. Ct. 2024); People v. Stubblefield, H048598 (Cal. Ct. App. 2024).
  30. People v. Johnson, H048633, H048722 (Cal. Ct. App. 2022).
  31. PENAL § 745(a)(3) and (a)(4).
  32. Id. § 745(h)(1).
  33. Id. § 745(b)(2).
  34. Id. § 745(a)(1) and (a)(2).
  35. People v. Stubblefield, H048598 at 26-27 (Cal. Ct. App. 2024).
  36. People v. Johnson, H048633, H048722 (Cal. Ct. App. 2022).
  37. Id.
  38. See Evan Kuluk, “Disparate Racial Impact of Discretionary Prosecutorial Charging Decisions in Gang-Related Murder Cases: Litigating the Racial Justice Act in People v. Windom,” Berkeley Journal of Criminal Law (2024).
  39. Court’s Order Re: 745(a)(3) Motion at 4, People v. Windom et al., No. 01001976380, (Cal. Ct. Contra Costa Cnty. fled May 23, 2023).
  40. Mosby v. Superior Court (2024) (Cal. Ct. App. 2024).
  41. Marnie Lowe argues that the RJA, in effectively establishing a “super-exclusionary rule” that jeopardizes all charges, convictions and sentences obtained by racially-biased policing, is designed to force substantial shifts in police behavior. Marnie Lowe, “Fruit of the Racist Tree: A Super-Exclusionary Rule for Racist Policing Under California’s Racial Justice Act,” Yale Law Journal (Jan. 2022).
  42. Paige v. California, 291 F.3d 1141 (9th Cir. 2002).
  43. PENAL § 745(a).
  44. See W. David Ball, “Modeling Meaning: Causal Inference Under the California Racial Justice Act,” (Jul. 24, 2024) (explaining how practitioners’ real-world models of criminal conduct and offender characteristics shape their inferences about racial disparities in similarly situated comparisons).
  45. See Paige v. California, 291 F.3d 1141 (9th Cir. 2002).
  46. See Joshua Grossman, Julian Nyarko & Sharad Goel, “Reconciling Legal and Empirical Conceptions of Disparate Impact: An Analysis of Police Stops Across California,” (2024) (analyzing disparate impact in California police stops under the state’s Racial and Identity Profiling Act and demonstrating how different statistical approaches to measuring disparities can yield varying insights into discriminatory practices).
  47. See Teamsters v. United States, 431 U.S. 324 (1977).
  48. See U.S. Department of Justice, Civil Rights Division, “Title VI Legal Manual” (accessed Mar. 5, 2025).
  49. See Fallis v. Kerr-McGee Corp., 944 F.2d 743 (10th Cir. 1991); Stout v. Potter, 276 F.3d 1118 (9th Cir. 2002).
  50. Id.
  51. See Smith v. Xerox Corp., 196 F.3d 358 (2d Cir. 1999).
  52. See U.S. Department of Justice, Civil Rights Division, “Title VI Legal Manual” (accessed Mar. 5, 2025).
  53. See Jonathan Feingold and Evelyn Carter, “Eyes Wide Open: What Social Science Can Tell Us About the Supreme Court’s Use of Social Science,” Northwestern University Law Review Online (2018); Mario L. Barnes and Erwin Chemerinsky, “What Can Brown Do for You?: Addressing McCleskey v. Kemp as a Flawed Standard for Measuring the Constitutionally Significant Risk of Race Bias,” Northwestern Univ. L. Rev. (2018).
  54. Grutter v. Bollinger, 539 U.S. 306, 378 (2003) (Rehnquist, C.J., dissenting).
  55. See David Alan Sklansky, “The Problems with Prosecutors,” Annual Review of Criminology (2018); Robin Olsen, Leigh Courtney, Chloe Warnberg, and Julie Samuels, “Collecting and Using Data for Prosecutorial Decisionmaking,” Urban Institute (Sep. 2018); Colleen V. Chien, W. David Ball, and William A. Sundstrom, “Proving Actionable Racial Disparity Under the California Racial Justice Act,” 75 Hastings L.J. 1 59 (2023).
  56. Robin Olsen, Leigh Courtney, Chloe Warnberg, and Julie Samuels, “Collecting and Using Data for Prosecutorial Decisionmaking,” Urban Institute (Sep. 2018).
  57. Committee on Revision of the Penal Code, “Staff Memorandum 2023-07: Prosecutorial Discretion, Plea Bargaining, and Related Matters,” 9 (Sep. 26, 2023).
  58. See Ram Subramanian, Léon Digard, Melvin Washington II, and Stephanie Sorage, “In the Shadows: A Review of the Research on Plea Bargaining,” Vera Institute of Justice (Sep. 2020); American Bar Association Criminal Justice Section, “Plea Bargain Task Force Report,” (2023).
  59. Id. Plea Bargain Task Force Report.
  60. Trout v. Lehman, 702 F.2d 1094 (D.C. Cir. 1983).
  61. A review of publicly available public records requests and local government meeting resolutions and minutes identified 15 California counties that have entered into contracts with Sicuro Data Analytics that, if fulfilled, could cost more than $6 million. The implications of concentrating this analytical work with a single firm remain unclear. On one hand, it may create a significant resource disparity between typically better-resourced prosecutors’ offices and more constrained public defender budgets. On the other hand, while Sicuro could potentially apply consistent analytical approaches across jurisdictions, there’s also the possibility that analyses will be tailored to the specific policy interests of each prosecutor’s office they serve, undermining the potential benefits of standardization.
  62. PENAL § 745(c)(1); See Daniel L. Rubinfeld and Joe S. Cecil, “Scientists as Experts Serving the Court,” Daedalus (Oct. 1, 2018).