Joseph L. Gastwirth and Qing Pan

Joseph L. Gastwirth is a professor of statistics and economics at The George Washington University and chair of the ASA Law and Justice Statistics Committee.

Qing Pan is an assistant professor of statistics at The George Washington University and a researcher at the biostatistics center.

This winter, the Supreme Court considered the case of Diapolis Smith, an African-American man convicted in 1993 of second-degree murder by an all-white jury in Kent County, Michigan. Smith requested a new trial, arguing that minorities were under-represented on venires in the county. The statistics employed in Berghuis v. Smith exemplify the increasingly important role statistical evidence plays in legal cases and the larger role statistical inference could play.

Two Supreme Court cases provide guidance on jury representation. In Taylor v. Louisiana, the Supreme Court stated that the Sixth Amendment guarantees a criminal defendant the right to a trial by an impartial jury drawn from a fair cross-section of the community. Subsequently, in Duren v. Missouri, the court stated three criteria a defendant must prove to establish a prima facie case: (1) a “distinct group” is identified, (2) the group was not fairly represented on the jury venires, and (3) the aspect of the jury selection process creating the “systematic exclusion” or under-representation of the group needs to be specified.

Statistical evidence is essential in the second step and often useful in the third. If the defendant’s evidence meets these criteria, the state must justify the practice in question by showing it advances a substantial societal interest (e.g., ensuring the care of young children).

In Berghuis v. Smith, the defendant relied on statistical evidence to satisfy the second and third criteria. African Americans formed 7.28% of Kent County’s age-eligible population, but 6% of the jury pool during the six-month period up to and including his trial. While the defendant suggested several factors caused the “systematic under-representation,” a major one was that potential jurors were first assigned to local courts, rather than the circuit courts hearing felony cases. A change in the system soon after the defendant’s trial allowed this argument to be checked. During the following year, African Americans formed 6.17% of the jury pool.

Courts have relied on three statistical approaches to assess under-representation (see Assessing Under-Representation). The Michigan Supreme Court noted that the U.S. Supreme Court had not specified the preferred approach; therefore, courts should consider the results of all three tests when the parties submit sufficient evidence.

That court gave the defendant the benefit of the doubt on the under-representation issue, but decided he had not shown that the assignment process caused the systematic under-representation. The 6th Circuit Court reversed the Michigan Supreme Court. Its opinion concluded that under-representation was demonstrated because the CD measure was -18% during the first period when the defendant’s trial was held and -34% the month of his trial. It accepted testimonial evidence that, at the time of the defendant’s trial, the process of sending minorities to local courts before the circuit courts led to their systematic under-representation on circuit courts.

Neither party submitted a standard deviation analysis, although an amicus brief to the U.S. Supreme Court from a group of academics that included statisticians did. That brief questioned the use of statistical testing in the context of fair representation because the number of individual jurors called for possible service can be so large that minimal deviations between the minority fraction of the pool and their fraction of the eligible population would be deemed significant. The amicus brief recommended that the court use the CD measure and that a CD = -15% or less be a general threshold, which should be modified to -25% when the minority group formed less than 10% of the jury-eligible population.

At the U.S. Supreme Court, the defendant argued that the CD was the appropriate measure of under-representation, while the state argued that the court should adopt a rule that an AD of at least 10% is required. This AD criterion has been criticized, as minorities forming less than 10% of the jury-eligible population could be totally excluded. The court noted that the magnitude of under-representation in this case was substantially less than in Duren v. Missouri, where the shortfall in female jurors met all three statistical criteria. The opinion reversed the 6th Circuit Court, which had relied solely on the CD measure. The U.S. Supreme Court reinstated the Michigan Supreme Court’s decision because it was far from “unreasonable” in applying the precedential cases that the standard federal courts should use under a 1996 law.

An important aspect of the statistical evidence that was not called to the attention of the courts is that the number of African Americans in the jury pool was an estimate. The reason for this is that the Michigan Department of Motor Vehicles’ database, from which potential jurors are selected, does not record racial identification. The defendant’s statistician used the residence pattern of members of the jury pool to estimate their race. This was done for each of the 112 census tracts by assuming the fraction of African Americans among the individuals from a tract equalled their fraction of the jury-eligible residents of that tract.

Pages: 1 2