The authors have data on the race, gender, and age of each member of the jury pool as well as each member of the ultimate jury. The authors also know the race and gender of the defendant and the charges. What the authors discover is that all white juries are 16% more likely to convict black defendants than white defendants but the presence of just a single black person in the jury pool equalizes conviction rates by race. The effect is large and remarkably it occurs even when the black person is not picked for the jury. The latter may not seem possible but the authors develop an elegant model of voir dire that shows how using up a veto on a black member of the pool shifts the characteristics of remaining pool members from which the lawyers must pick; that is, a diverse jury pool can make for a more “ideologically” balanced jury even when the jury is not racially balanced.
--Alex Tabarrok, Marginal Revolution, on the elusive goal of color-blindness