Scene: Athens, Georgia, a courtroom. A pool of about fifty potential jurors, including me, is being questioned by the defense attorney. Next to him sits his client, a young black male who has been accused of raping a white woman.

Attorney: Now, by show of hands, do any of you think that members of certain groups in our society are more likely to commit violent crimes than members of other groups?

Me, thinking to myself: Hmm, what would a “no” answer imply? That the prison population is a perfect cross-section of the general population? Clearly not the case. That any demographic over-representation in the prison population is entirely due to bias in arrests and convictions? Hard to believe.

Here are the sort of numbers that were in the back of my mind.

(Source here. The most recent Bureau of Justice Criminal Offenders Statistics are here.)

I raise my hand. (Yes, I am saying, what I know indicates that not every group commits violent crimes at the same rate.) No other hands go up.

Attorney, to me: Okay, which groups?

Me: Well, males, especially young males. The economically disadvantaged. Ummm ... (do I really have the nerve to say this next thing?) … blacks.

[Note: I’ve since heard that “black” may drop into statistical insignificance if you control for “raised in a single-parent household”; the two regressors are highly correlated among Americans. Anyone know whether that’s true?]

[Second note: I did not have a chance to explain to the defense attorney that my answers to his question were irrelevant to how I as a juror would decide in the case. The demographics of the criminal population are only relevant to predicting in a random case, which means independent of the evidence in the case at hand, what the criminal might look like. If compelled to serve on the jury, I intended to listen to the evidence. The question to be tried was whether this particular individual was guilty of the crime with which he was accused in this particular case. The relative frequency with which other members of his demographic commit violent crimes has no bearing.]

Attorney: Okay, does anyone else agree with that?
(Three people raise their hands.) Do you want to add anything?

One of the three: Also, people with less education.

None of the four of us were selected for the jury. What we said was a statistical answer to a statistical question, not an expression of racism. Nonetheless the defense attorney may have rationally excluded us, figuring that those who didn’t agree with us were better indicating that they would not be biased against a black defendant. They were indicating that they would not volunteer to say anything that might seem unsympathetic to the defendant. Any (non-dissembling) anti-black racist in the pool would raise his hand (for the wrong reason); anybody who thinks that “over-representation of blacks in the prison population is entirely due to bias in arrests and convictions” would not raise his hand.

Postscript: The defendant was acquitted. Based on what I later read in the local newspaper about the flimsiness of the case against him, I’d have voted to acquit too.