Rivera v. Illinois and “Reverse Batson”

linkedin
Share on Google+
Share on Reddit
Share on Tumblr
Download PDF

In Batson v. Kentucky, 476 U.S. 79 (1986), the Supreme Court held that prosecutors could not exercise peremptory challenges based on race. In Georgia v. McCollum, 505 U.S. 42 (1992), the Court extended the same rule to defendants. (Sex discrimination is likewise prohibited, under J.E.B. v. Alabama ex rel T.B., 511 U.S. 127 (1994).) Yet while defendants regularly make Batson challenges to prosecutors’ strikes, it appears that prosecutors rarely return the favor. In fact, there is only one reported appellate case in which a North Carolina prosecutor successfully challenged a defendant’s use of peremptory challenges: State v. Cofield, 129 N.C. App. 268 (1998).

Perhaps prosecutors have been reluctant to invoke McCollum because of uncertainty about the consequences of success. Clearly, if a McCollum challenge succeeds, the defendant is convicted, and the trial judge’s ruling is upheld on appeal, the conviction stands. But what if a McCollum challenge succeeds, the defendant is convicted, and the appellate courts conclude that the trial judge erred in granting the challenge? Prosecutors — and judges — may have assumed that the appellate courts would find so-called structural error, mandating automatic reversal. Although no North Carolina appellate case so held, the United States Supreme Court stated in dicta in Swain v. Alabama, 380 U. S. 202, 219 (1965), that “[t]he denial or impairment of the right [to exercise peremptory challenges] is reversible error without a showing of prejudice.” Fear of structural error may have deterred prosecutors from contesting defendants’ questionable use of peremptory strikes.

McCollum challenges may become more common as a result of Rivera v. Illinois, __ U.S. __ (2009), available here, a case just decided by a unanimous United States Supreme Court. In Rivera, the trial judge sua sponte invoked McCollum to question a defendant’s peremptory challenge of a black female juror. After hearing defense counsel’s explanation for the strike, the judge ruled that the defendant was attempting to exercise the peremptory challenge in a discriminatory fashion, and, because the juror in question was not subject to challenge for cause, required that she be seated over the defendant’s objection. She ultimately served as the foreperson of the jury that convicted the defendant.

The defendant appealed, arguing that his challenge was not based on the juror’s race or sex, and that the trial judge’s rejection of the defendant’s peremptory challenge was structural error that required a new trial. The state appellate court agreed with the first part of the defendant’s argument, finding no evidence that the challenge was discriminatory, but disagreed with the second part. It ruled that, because no juror who was actually seated was biased against the defendant, the defendant was not deprived of his constitutional right to a fair an impartial jury. He was, of course, deprived of his state statutory right to use his peremptory challenges as he saw fit, but the court ruled that the error was harmless beyond a reasonable doubt in light of the strength of the case against him.

The United States Supreme Court granted certiorari to resolve a split among state supreme courts about whether the erroneous denial of a peremptory challenge automatically requires a new trial. It affirmed the state supreme court, finding that “the loss of a peremptory challenge due to a state court’s good-faith error is not a matter of federal constitutional concern. Rather, it is a matter for the State to address under its own laws.” Specifically, it held that because no member of the defendant’s jury was subject to removal for cause, he was not deprived of his right to an unbiased jury; and although he was subjected to a violation of state law, it was not such a fundamental insult that it rose to the level of a due process violation.

Does this mean that the North Carolina appellate courts will review successful McCollum challenges for harmless error, as the Illnois courts did in Rivera? That result isn’t strictly required by Rivera, since the federal courts can’t tell state courts how to review errors of state law, but it certainly seems likely. And that may give prosecutors a little more comfort in invoking McCollum. That appears to have been one of the Supreme Court ‘s goals, as it explained its decision in part by noting that “[t]o hold that a one-time, good-faith misapplication of Batson violates due process would likely discourage trial courts and prosecutors from policing a criminal defendant’s discriminatory use of peremptory challenges.”

3 comments on “Rivera v. Illinois and “Reverse Batson”

  1. One recent case that has some language dealing with this issue is State v. Williams, ___ N.C. App. ___, 662 S.E.2d 397 (July 18, 2008), in which the defendant on appeal argued that he had reached an absolute impasse with his trial counsel over the used of his peremptory challenge. The Court of Appeals held that their was no absolute impasse but went on to say that, even if there were such an impasse, the defendant’s trial counsel could not have lawfully complied with the defendant’s wishes with respect to the exercise of his peremptory challenges because the defendant admitted that he wanted to strike white jurors until the jury was evenly divided “half and half” between whites and blacks.

    NB: The defendant’s appellate counsel filed a notice of appeal (constitutional question) and a PDR in the NC Supreme Court in the Williams case, and both filings are still pending at this time.

  2. Bah, forgive the typos above — first sentence “used s/b “use” and “challenge” s/b “challenges”, second sentence “their” s/b “there”.

    I hate when that happens.

  3. […] that defense lawyers raise such claims at least as often as prosecutors do. (Jeff suggests in this post from 2009 that prosecutors may have been reluctant to raise reverse Batson claims because if they […]

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.