PERSONAL PERSPECTIVE : Clouding the Simpson DNA Issue: Let the Numbers Games Commence

Determining whether or not the DNA profile of blood samples taken from the crime scene matches that of O.J. Simpson is only half the battle. Then the numbers games can begin.

A DNA test is, in essence, an examination of certain physical traits. In this sense, the forensic expert’s testimony that the blood stain at the crime scene matches that of the suspect is akin to an eyewitness’ statement that a defendant “looks like” the person seen committing the crime. But the analogy will only stretch so far.

A DNA match means similarity for each of a handful of specific traits. It is as if an eyewitness asserts that the faces of suspect and of perpetrator share a certain mole and color of eyes but that the witness is unable to observe skin color, shape of nose or texture of eyebrows and many other features that add up to an impression of facial identity. Consequently, the format of DNA evidence presented in court will be unfamiliar to jurors.

An ordinary eyewitness might claim “100% positively” that the suspect is the perpetrator of the crime. Such a person is simple to understand though not altogether believable. Jurors may discount the witness’ claim of certitude--to say, 95%--in light of their experience in judging people and other evidence presented at trial.

The forensic expert presenting DNA evidence, on the other hand, may be 100% believable in asserting that the crime sample and reference (suspect) sample match (“look alike”) with respect to certain traits. But the expert admits “doubt” when he says that DNA profiles from different sources may have a small chance (“match probability”) of looking alike. It is not obvious to a juror what relationship, if any, the expert’s “doubt” has to an eyewitness’ credibility. Our intuition doesn’t easily digest numbers.

The most frequently repeated confusion concerns the meaning of “match probability.” For example, a recent news story reported that some preliminary blood analysis in the Simpson case indicated a match probability of 1/100--meaning, two samples share a blood type that only one person in 100 has. In elaborating on the meaning of the match, the story explained that the chances of the two samples coming from different sources was 1 in 100.

The two statements may sound similar but they do not assert the same thing. Match probability assumes different sources, then asks: What is the chance of a match? This is only one step toward answering the critical and converse question, which can only be answered by a jury: Assuming two blood samples match, what is the chance that they came from different people?

Match probability is a statistical fact about a particular blood type, determined by surveying the population. By contrast, the chance that the samples come from different sources depends also on what is known about the circumstances. For example, if there is good reason to believe that the suspect was a continent away at the time the crime stain was made, that does affect the odds.

Or consider the math attached to the two DNA technologies--PCR and RFLP--in use for forensic analysis. PCR is newer, less firmly established and generally less discriminating than RFLP tests.

But PCR has some advantages, the most salient of which is that it can be applied to much smaller blood samples. Therefore, the most critical evidence in the Simpson case may well be match probabilities based on PCR typing. You can be sure that both prosecution and defense are now considering the following scenario.

Suppose a lab performs a battery of four or five different PCR tests on a critical blood sample. Assume, also, that some crime-scene DNA sample turns out to match Simpson’s. There will be two possible explanations: a) the DNA, in fact, comes from Simpson’s blood or, b) it comes from someone who coincidentally shares Simpson’s DNA profile.

To assist the jury in weighing these possibilities, the prosecution’s DNA expert will quantify the coincidence as, say, 1 in 10,000 if the “someone” is a random, unrelated person. It will be intriguing to watch how the defense subtly stresses the assumption “unrelated” and tactfully manages to insinuate the unspeakable but beneficial hypothesis that the crime-scene DNA might come from a relative of Simpson’s.

“Unrelated, you say?” asks the defense. “Then the odds would change if we considered a relative?”

“Of course,” replies the forensic scientist. “Relatives automatically have some DNA traits in common, so there is less remaining to explain as coincidence.”

“Purely hypothetically then, and not for a moment suggesting that a Simpson relative, say his son, could have committed anything so heinous as murder--but just for the sake of illustration, what would be the chance of a match between him and his father?”

“In this sample from the crime scene were left by Simpson’s son, parent, even sibling, there would be a 1 in 200 chance that Simpson’s would also match it.”

The results from Cellmark Diagnostics will be based on RFLP analysis. When the outcome is known, it will either be a mismatch (everybody goes home) or a match probability in the range of one in 10 million to one in a billion.

These are intimidating numbers, but the defense has many ways to shave them--excluding some tests, disputing the statistics, challenging the statistical methodology. The matching odds--whether 200 to 1 or 10,000 to 1--are obtained by “cascading” (i.e., multiplying) inferences from a handful of individual tests each with very modest evidentiary power--odds of perhaps only 3 to 1.

Indeed, in two recent appellate decisions in Northern California, the courts disallowed multiplication, holding that such a calculation fails to meet the Kelly-Frye standard of uniformly accepted practice in the scientific community. The impression of scientific dissent that influenced the rulings was due to the heterodoxy expressed in a single scientific paper, notwithstanding voluminous data, as well as logic, supporting the majority view.

If such a ruling were repeated in the Simpson case, the weight of the DNA evidence would appear to be severely limited. It is a fair bet that telephones of the dissenting scholars have already begun to ring.