Statistical measures and methods for assessing the representativeness of juries: a reanalysis of the data in Berghuis v. Smith

Abstract

To assess the fairness of a jury system, courts compare data on the race-gender-ethnic composition of a number of jury venires (pools) to that of the jury-eligible population of the jurisdiction. The recent U.S. Supreme Court opinion in the Berghuis v. Smith case discussed the criteria a defendant's evidence must meet to establish a prima facie case that the jury pools were not representative. The decision relied on the three criteria set previously in Duren v. Missouri: name a distinct subgroup of the population, demonstrate their underrepresentation with statistics and identify the aspect of the jury system, which produced the disparity. In addition to formal statistical testing, courts have considered the absolute and comparative differences between the percentages of minorities in the age-eligible population and on the jury venires. In Berghuis, data for the period when the defendant's trial was held indicate that 6.03% of the potential jurors in the circuit (felony) courts were African American but they formed 7.28% of the age-eligible residents of the County. The 1.25% absolute difference seems small. The comparative difference or percentage shortfall is 18%, suggesting underrepresentation. The defendant argued that the disparity in the circuit courts arose because, at the time of his trial, jurors were first assigned to local courts and African Americans were concentrated in Grand Rapids. Shortly after the trial, the County changed the system, however, the comparative disparity (CD) only declined by 3%. This small change failed to convince the Court that the juror assignment system caused the underrepresentation. The available data in Berghuis v. Smith differed from most jury discrimination cases as the racial identification of the venire members were unavailable. The defendant's expert estimated the number of African Americans on the venires from their addresses and compared it to their expected number derived from their fraction of the county's age-eligible population. He did not test whether the difference was statistically significant. The two analyses available to the Court treated the estimated number of minority jurors on the venires as though it was the actual number, which is statistically incorrect. This article derives the appropriate standard deviation analysis. Applying it to the data in Berghuis, it is shown that there was a statistically significant disparity between the percentages of African Americans on the venires and in the age-eligible population. On the other hand, the difference between the CDs in the two periods was not near significance. Thus, the first two Duren criteria were satisfied but the third was not, supporting the Court's decision. Language: en