Nearly 20 years ago, The Dallas Morning News reported that Dallas County prosecutors were excluding nine out of 10 blacks from jury service. So when the U.S. Supreme Court agreed to review a Texas death-penalty case that alleged jury discrimination, we thought it was time to see whether anything had changed.
After two years of research, multiple court orders and regression models, we had our answer: race still played a key role in who serves on a jury.
But because the trend was not as blatant as it was in 1986 - when reporters Steve McGonigle and Ed Timms found that four out of five juries were all white - the story required a more sophisticated statistical tool.
We used logistic regression, which computes the relationship between variables, when the dependent (or outcome) variable is dichotomous - meaning it has two possibilities for the value.