Journalists should be required to take a statistics course. Maybe then we wouldn’t be inundated daily with so many absurd sports stats, political poll results, and crime “urban legends.” Fortunately, somewhere in history, lawyers and judges learned that statistics aren’t evidence and should not be presented to juries.

In the first Phil Spector trial (2007), at least one of his expert witnesses came very close to telling the jury statistics showed that gunshots to the mouth were almost always self-inflicted, but prosecutor Alan Jackson quickly pounced on such statistics as inappropriate testimony.

In Scott Peterson’s trial (2004), the public knew only what journalists told them about the evidence, since no cameras were permitted. So, I don’t know whether statistics were presented to the jury. But journalists repeatedly claimed that “the leading cause of death among pregnant women is homicide by the spouse or significant other.” This was a false statistic, but it condemned Peterson in the public’s eyes (and I assume it made its way into the minds of the jurors, as well).

Statistics are mathematical computations based on data. Like all computations they are subject to error both in the way they are computed and in the data on which they are based (“garbage in, garbage out”). Statistics are also prone to misinterpretation (in science, as well as in journalism). Statistics don’t prove whether or not Lana Clarkson killed herself or whether or not Scott Peterson killed his wife.

Using statistics the way journalists use them is tantamount to relying on stereotypes to identify criminals. The media make gross assumptions about criminals: “mothers kill their children,” “husbands kill their wives,” “gangsters are drive-by killers,” and so on. Statistics may (or may not) exist to support these stereotypes, but that does not mean in every case the killer must fit the stereotype. Yes, Andrea Yates killed her children, but the Ramseys did not kill Jonbenet. Yes, husbands kill their wives, but sometimes a man like Fred Cooper kills another man’s wife who witnesses him killing her husband. Yes, gangsters are often drive-by killers, but sometimes the drive-by killer is a sociopathic Colorado youth pretending he’s Al Pacino in Scarface.

The cops rely extensively on stereotypes to identify criminals. It’s not smart, but it’s understandable—like looking for your lost car keys under the nearest lamp post, even though you know you dropped them several feet away in the dark parking lot. But false stereotypes also often lead the cops in the wrong direction (as in the tragic case of Jessica Lunsford).

Criminal defense attorneys are vigilant against the use of stereotypes in law enforcement—even the use of “criminal profiling” (a questionable investigative tool, IMHO). “Racial profiling” is rightly deemed the equivalent of stereotyping and racism.

The courts exclude statistical evidence, because it’s prejudicial, not probative.

Jurors are human, though. Jurors hold stereotypes. Voir dire is intended to expose juror’s stereotypes, but from time to time an innocent person is convicted because he or she fits a criminal stereotype.

When statistics are misunderstood and abused, the result is stereotyping and urban legends. And even when statistics are understood and properly used, they only tell you about general principles, not specific crimes. Yes, gunshot wounds to the mouth are “usually” self-inflicted; but occasionally a murderer shoots a person in the mouth. Occasionally strange things do happen.

To learn about statistics, I recommend Professor Michael Starbird’s The Teaching Companies course, “Meaning from Data: Statistics Made Clear.” It’s fun as well as informative.

In Lecture 13, “Law—You’re the Jury,” Prof. Starbird shows several misuses of statistics that either have come up in court or could easily come up if the presiding judge isn’t careful.

• Hit-and-run accident: A witness claims his identification of the offending vehicle as a blue taxi is better than 50% accurate.
• Firing an employee: A company uses random drug testing, which a fired employee claims is prone to false positive results.
• College admissions: A woman claims that a college’s admissions policies are discriminatory based on the percentage of women whose applications are accepted versus the percentage of men whose applications are accepted.
• O.J. Simpson murder trial: Johnny Cochrane’s closing argument claimed that O.J. was an unlikely killer because only 1 in 1,000 wife-beaters kill their wives.
• Another case involves an interesting misuse of DNA statistics, similar to the misuse in the Joshua Rosa trial.