California’s Racial Justice Act Goes to Court

In a quartet of cases, the California Supreme Court last week made its first rulings interpreting the state’s Racial Justice Act (RJA), a landmark statute designed to eradicate racial bias from criminal prosecutions. Among its rulings, the court overturned a death sentence imposed after a prosecutor compared the defendant, a Black man, to a “Bengal tiger.” But otherwise, two justices argued, the court narrowed the law’s protections — raising the bar for RJA violations and allowing “harmless” violations to go unremedied — in ways contrary to the statute’s text and broad remedial purpose.

California’s RJA, enacted in 2020, is the only law of its kind in the country. In enacting the law, the legislature declared its intent was to eliminate bias based on race, ethnicity, or national origin “in any form or amount, at any stage of a criminal trial.” Similarly, lawmakers were acting on “growing awareness that no degree or amount of racial bias is tolerable in a fair and just criminal justice system,” and that systemic or implicit bias can undermine fairness to the same extent as explicit bias or intentional discrimination. Among its provisions, the RJA prohibits language during criminal trials — whether delivered by a prosecutor, judge, witness, or defense counsel — that “implicitly appeals to racial bias,” even if that appeal was not “purposeful.” With this expansive mandate, the RJA’s goal is not to punish racists or even identify cases where discrimination changed the outcome; it is to eliminate racial bias from the legal system, whatever its form and whatever its role in a particular case.

To that end, the statute provides broad, mandatory remedies for violations: When a court finds a violation on appeal, it “shall vacate the conviction and sentence” (or just the sentence, if the violation occurred at a sentencing proceeding). After violations in capital cases, the statute says that “the defendant shall not be eligible for the death penalty.”

In addition to confronting the insidious harm of implicit bias, the RJA recognized the failure of courts to address it. When enforcing rights to a fair trial, the U.S. Supreme Court insists on proving intentional discrimination before affording a judicial remedy. This is true in jury selection, where Batson v. Kentucky, decided in 1986, prohibits intentionally excluding potential jurors based on race. And it is true in sentencing, where, under McCleksey v. Kemp, decided one year later, racial disparities are insufficient to establish that death sentences are improperly based on race. With the RJA, the California legislature explicitly rejected McCleskey, declaring that “we can no longer accept racial discrimination and racial disparities as inevitable in our criminal justice system.”

In the cases decided last week, the California Supreme Court for the first time considered the RJA’s scope, particularly in death penalty cases with trials that predate the law. In the case of Anthony Bankston, the court found a violation and overturned his death sentence because the prosecutor compared him to a Bengal tiger at the zoo. Prosecutors told jurors a well-worn tale of a journalist first observing a Bengal tiger at the zoo, “real lethargic,” before traveling to see one in the wild. There, the story went, he finds a tiger “all flexed out, he sees the claws out, he sees the fangs … he hears the growl.” The jury should think of the defendant the same way, the prosecutor implied, appearing one way in the courtroom and another when he’s not in view of the jury.

The Bengal tiger story was historically so common in California prosecutions that the legislature used it as a specific example of an RJA violation. The state attorney general also conceded that the story improperly appealed to racial biases and evoked the sort of comparisons to hyper-predatory animals that have long been used to dehumanize and criminalize Black men. As Justice Goodwin Liu wrote in a concurring opinion, this RJA violation was “rather obvious.” 

Beyond this finding, though, Liu and (especially) Justice Kelli Evans criticized the court for unilaterally narrowing the RJA, both as to what constitutes a violation and whether certain violations might be “harmless” and therefore left unremedied.

As to the scope of RJA violations (i.e., the sort of language that improperly appeals to bias), the court declined to find violations based on other comparisons between animals and nonwhite defendants — despite legislative findings that “because use of animal imagery is historically associated with racism, use of animal imagery in reference to a defendant is racially discriminatory and should not be permitted in our court system.”

In one case, the prosecutor described a Mexican defendant as “inhuman” and someone who could be called “an animal, but I would not insult animals.” In another case, the prosecutor referred to a Black defendant as a “predator” and “wolf in sheep’s clothing.” Combined with Bankston, these rulings “hold, bizarrely, that the RJA is violated when a Black defendant is compared to a violent, predatory tiger but not when compared to a violent, predatory wolf,” Liu wrote. According to the majority, the distinctions lie in the manner each reference was made. “But the Legislature did not intend that courts narrow the RJA’s prohibition[s] … with such qualifiers,” Liu went on, and “comparing a Black defendant to a violent, predatory animal, of whatever species and in whatever manner, carries significant risks of dehumanization, moral exclusion, and unfairly harsh punishment.”

The court also declined to find violations based on cultural and nationality stereotypes. In one case the defendant’s immigration status was a recurring theme of his penalty trial, including expert testimony that being “illegal” was a risk factor for abusing children. In another case with a defendant from Cambodia, “the prosecutor,” Liu wrote, “invoked cultural scripts to set the jury’s expectations for how Cambodian immigrants should behave and then leveraged those stereotypes to cast [the defendant] as deviant for failing to conform to expectations associated with his ethnicity.” Such “gratuitous references to nationality, race, or immigration status,” Liu said, are bound to activate unconscious biases and are therefore straightforward violations of the RJA.

Writing separately, both Liu and Evans said the court undercut the RJA by ignoring appeals to racial bias if they were made in service of a message that the jury could properly consider. “I fear this point cannot be emphasized enough,” Evans wrote, adding that, “An appeal to racial bias is not excused or mitigated merely because it is part of an evident effort to comment on matters fairly presented to the jury for its consideration.” The RJA, “after all, is not limited to gratuitously tossed-off epithets that are not a fair comment on the evidence,” she said. “That type of language was already misconduct before the RJA was enacted.”

In addition to narrowly defining violations, the court diluted two other core features of the RJA: its automatic remedies and its bar on death sentences after a violation. On both issues, the court invoked the doctrine of “constitutional avoidance,” basing its narrow statutory reading on the need to avoid difficult constitutional questions.

First, the court found a “harmless error” standard for cases tried to judgment before the RJA took effect in 2021 — and on this point only Evans disagreed. With this ruling, the court wrote, “The use of language that implicitly appeals to racial bias is not grounds for reversal if [the state] can show that the use of such language was harmless beyond a reasonable doubt.”

The court’s analysis on this point involves a complex mix of statutory and constitutional interpretation that ultimately misses the forest for the trees. The RJA indeed limits relief for “petitions” filed in cases before the RJA was enacted if “the state proves beyond a reasonable doubt that the violation did not contribute to the judgment.” The term “petition” generally refers to habeas corpus petitions filed after appeals are exhausted and judgment is final, Evans pointed out, whereas the four cases before the court were still on direct appeal. But reading “petitions” as excluding these cases, the majority noted, would require their automatic reversal under the RJA — in potential conflict with a 1911 state constitutional amendment providing that errors in California criminal trials warrant relief if “the court” finds there was a “miscarriage of justice.” Finding ambiguity as to the meaning of “petitions,” the court opted to apply harmless error review even to cases on direct appeal. By preserving the court’s role in assessing whether relief is warranted, the court avoided the “serious constitutional questions” presented by the 1911 amendment.

The RJA’s primary author, Assemblymember Ash Kalra, cautioned the court against this outcome in an amicus brief: “To interpret the Racial Justice Act as only allowing relief if the defendant was prejudiced would be antithetical to the Legislature’s purpose because it would allow some threshold level of racial bias to be acceptable,” he wrote. Evans echoed this in her opinion. “The RJA as construed by my colleagues is substantially diluted,” she wrote. Now “the question will no longer be Was the language used an appeal to racial bias? but will instead be How much racial bias is too much? Such an approach seems likely to bring the judicial system into disrepute.”

The court similarly weakened the RJA’s mandate that, after a violation, “the defendant shall not be eligible for the death penalty.” Instead, it held that clause applies only to the proceeding in which the violation took place, such that Bankston remains eligible for the death penalty at resentencing. The court also held that its harmless error standard applies to death sentences, and so courts can uphold a death sentence if the RJA violation was harmless beyond a reasonable doubt. This effectively adds a third step between finding a violation and removing death eligibility. Now, after finding a violation, the question is whether it was harmless.

Again, the court raised “serious constitutional concerns” based on a California ballot initiative. This time, it was the 1978 “Briggs Initiative” which establishes criteria for death penalty eligibility and that the legislature cannot amend. Reading the RJA to bar death sentences at retrial could, the court said, be taken as an improper amendment to this death penalty statute. And again, Evans argued that “the canon of constitutional avoidance [does not] grant us the authority to rewrite this important criminal justice reform statute.” She added that “nowhere in the RJA’s text did the Legislature suggest there was an additional step between the finding of a violation … and the obligation to provide an effective remedy.”

Ultimately, Evans wrote, the court was contorting itself to avoid “too much justice.” And by doing so, it continues to do what courts have always done: minimize discrimination and bias at every turn so as to avoid meddling in criminal prosecutions. But the RJA was designed precisely to stop that practice — to overcome courts’ institutional reluctance and mandate intervention whenever and however racial bias appears. “The RJA, in other words, is a directive to courts to change,” Evans wrote, and the “change the Legislature mandated requires courts to be willing to recognize racism when it occurs and to provide an effective remedy each and every time.”

Kyle C. Barry is executive director of the State Law Research Initiative.