Len Niehoff is Professor from Practice at the University of Michigan Law School, where he teaches courses in civil procedure, ethics, evidence, First Amendment, law & theology, and media law. He writes regularly in all of these fields. He is also Of Counsel to the Honigman law firm. The opinions expressed here are his own.

Tuesday, September 16, 2014

A Parable on Race, Statistically Rendered

On June 18, 1964, Juanita Brooks was walking home through an alley.  She used a cane and was pulling behind her a wicker carryall that contained her purse and the groceries she had just purchased.  Someone rushed up behind her, knocked her down, and grabbed her purse.  Ms. Brooks saw a young woman running from the scene.  Other witnesses saw a woman running as well, and observed as she jumped into a car that sped off.

This may seem like the most mundane of crimes, but it led to a decision from the Supreme Court of California that appears in major Evidence textbooks.  The fiftieth anniversary of these events offers an appropriate occasion to pause and ponder the case and what it may be able to teach us today.

The case is People v. Collins, 68 Cal. 2d 319 (1968) and finds its way into standard Evidence texts because of the uncommon way in which a prosecutor tried to solve a fairly common problem.  The prosecutor's dilemma was that the witnesses to the crime provided conflicting testimony that made the identification of the defendants as the perpetrators less than certain.  So the prosecutor came up with a strategy he thought might help his case. 

The witnesses seemed to agree that the woman was Caucasian and had blondish hair in a ponytail, the driver of the car was a Black male with a mustache and beard, and the car was medium-to-large and yellow.  The alleged perpetrators had these same characteristics.  Defendant Janet Collins was a white woman with blondish hair that she often wore in a ponytail.  Defendant Malcolm Collins was an African American male and, though he had no beard at the time of trial, he could of course have shaved it off.  They owned a yellow Lincoln.  The prosecutor thought the probabilities were with him: what are the odds, he wondered, that I have the wrong defendants?

To try to convey this to the jury, the prosecutor called an expert statistician to testify.  The statistician explained the "product rule," which holds that the probability of the joint occurrence of a number of mutually independent events is equal to the product of the individual probabilities that each event will occur.  Thus, the probability of rolling two 2's successively on a die is 1/36: that is, 1/6 x 1/6.

The prosecutor then gave the statistician a number of individual probabilities; the odds of an automobile being partly yellow are 1 in 10, the odds that a man has a mustache are 1 in 4, the odds that a woman has blonde hair are 1 in 3, the odds that a woman wears her hair in a ponytail are 1 in 10, and so on.  The statistician multiplied all of these together and arrived at the conclusion that there was only 1 chance in 12 million that any couple possessed the distinctive characteristics of the defendants--whom the jury promptly found guilty.

Unfortunately, there were numerous errors in this methodology.

First, the prosecutor simply made up the individual probabilities that he gave the statistician.  There was no reason to believe that any of them were correct, let alone all of them.

Second, the product rule only applies where the events being multiplied are mutually independent.  But many of the facts used in this calculation overlapped--for example, some blonde women have ponytails.

Based on these and other flaws in the calculation, the Supreme Court of California held that it was error to admit the evidence.

Textbooks include Collins because it provides excellent fodder for a discussion of the ways in which unreliable mathematical evidence can unfairly prejudice a jury.  The California Supreme Court repeatedly notes the persuasive power of mathematics--"a veritable sorcerer in our computerized society" that can "cast a spell" over a jury, a discipline rich in the "mystique" of certainty.  It is easy to understand--and fun to discuss--how the use of such phoney mathematics could have confused and misled the jury in violation of Federal Rule of Evidence 403.

Still, stopping our analysis there seems unsatisfying.  After all, one might reasonably question whether the jury placed much weight on the 1 in 12 million number.  In his closing argument, the prosecutor acknowledged that the individual statistics were just his "estimates" and he invited the jury to substitute their own if they preferred.  It isn't clear that the jury would have been bewitched by the ostensible certainty of a mathematical formula under these circumstances.  In addition, Collins tells us little about how a court would probably approach this issue today.  Most courts today would bar such expert testimony through the swift and simple application of Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).

In my judgment, Collins was and remains interesting and relevant for a different reason related to a different potential source of jury prejudice.  That issue is revealed by a close look at a footnote in the Collins decision, where the California Supreme Court recounts the various odds that the prosecutor presented to the jury.  In that table, one "probability" clearly stands out, dramatically different from all the others: the odds that you would have an interracial couple in a car, the prosecutor submitted, were 1 in 1000.

Like all the others, this statistic had no foundation in fact.  But it would have left an impression with the jurors that the other statistics did not: these defendants were social outliers; each had done the highly unusual thing of marrying someone from another race; perhaps they felt comfortable breaking other "rules" as well.  In 1964--years before the Supreme Court would, in Loving v. Virginia, 388 U.S. 1 (1967), strike down laws banning interracial marriage--this would have been a subtext at the trial.  The 1 in 1000 statistic would have fed the narrative. 

Indeed, the statistic subtly underscores the message by referring to the odds of an interracial "couple" in a car.  But, of course, the witnesses did not know whether they were seeing a "couple."  They just knew they were seeing a male and a female in the same vehicle.

I think that Collins is moderately interesting, and of limited continuing relevance, as a "statistics-gone-bad" case.  But I think it is profoundly intriguing and persistently important as a case about race.  In Collins, race is wholly invisible, and also completely present.  It is, to appropriate Gilbert Ryle's wonderful phrase, the "ghost in the machine," the primary driver of the potential prejudice of the evidence and the court's anxieties, and yet spectral, sublimated, and suppressed.

Indeed, one of the few references the court makes to race has a stunning irony to it: the court entertains the possibility that eyewitnesses may have been mistaken because the woman in the car might have been a light-skinned "negress."  In other words, it might turn out that this wasn't an interracial couple after all.

In my view, Collins remains engaging and significant not as a case, but as a parable.  As with all parables, it yields more than one lesson.  But, surely, it offers us this lesson if none other: the most troublesome dialogues about race may be those that we will not even see or confess we are having.

No comments:

Post a Comment