Judges Gone Wild

CAUTION: This post contains plots corrected in Judges Gone Wild, Revised.

As I work on the summary chapter for my monograph Regarding the Rate of Wrongful Conviction, I keep finding new stories in all the data I've compiled and created. I frequently find myself in awe of what is unfolding before me. I'm still wrestling with the significance of the figure I displayed in Burden of Proof As a Legal Fiction. The one I am about to show you is equally revealing and equally troubling.

I wanted a way to display the odds an innocent person faces when tried for a crime. I knew the numbers Bruce Spencer calculated from the NCSC judge-jury agreement data. (He only had one number for the judge and one number for the jury, since the NCSC data were not broken down by crime.) I also had a sense of the numbers I had for the jury on a crime-by-crime basis. I had derived them from the Kaven-Zeisel judge-jury agreement data. My jury numbers (plural) corresponded reasonably well to Spencer's jury number (singular).

I only recently (as in today) got around to deriving my judge numbers from the Kalven-Zeisel data.

Whoa!

Here you go. Click to enlarge. We'll talk after you pick yourself up off the floor.

Professor of Statistics Bruce Spencer calculated you stand a 25% chance of being convicted by a jury if you are innocent. That value corresponds to the bottom of the green horizontal bar. My calculations show that the value is sensitive to the type of crime, and is slightly less than Spencer predicted. My values are indicated by the bottom of the brown vertical bars. They range from a high of 32% for drugs to a low of 14% for forcible rape. For a substantial majority of the trials, excluding drugs, the number is around 16%. That's one chance in six. That's bad enough, but better than the one chance in four predicted by Spencer.

Bruce Spencer and I, however, disagree on the risk an innocent party faces at the hands of a judge. Spencer predicts that judges (during bench trials) convict 37% of the innocents who come before them. That's horrible, but not nearly as horrible as I calculate. My calculations indicate that those people wrongfully charged with murder are convicted in 73% of the bench trials. Instead of slightly more than one time out of three, I calculate they are convicted three times out of four.

I'm discouraged.

The best of a bad crime-category lot is forcible rape. I calculate that judges convict "only" 35% of those people wrongfully charged with and tried for forcible rape. The judges' performance for every other crime category is worse.

Worse than a 35% chance of convicting an innocent person.

Worse.

As bad as a 73% chance of convicting an innocent person.

ERRATA: This post contains plots corrected in Judges Gone Wild, Revised.