- The Washington Times - Wednesday, February 18, 2004

The use of DNA analysis in criminal investigations is well known, but not always well understood. The chemistry is complex, and the terms used can be daunting: Restriction-fragment length polymorphism and polymerase chain reaction, for example.

The chemistry usually doesn’t matter in the outcome of trials. The technicians understand it, and it works. What does matter is that people do not always understand what the evidence means. The combination of impressive terminology and large numbers (“one chance in 150,000 that…”) can produce an air of infallibility.

A particular form of misinterpretation is common enough that it has a name, “The Prosecutor’s Fallacy.” The air of certainty sometimes associated with DNA analysis is justified when it is used to prove a negative. If the DNA in the blood at the scene does not match the DNA in John’s blood, then the blood isn’t John’s. But it doesn’t work so cleanly the other way. If the DNA at the scene does match John’s, it is not certain that it is his blood.

Suppose that a woman is raped. DNA analysis of the recovered semen reveals that one male in 100,000 would have DNA matching the sample. One million males live in the city. Suppose that it is discovered that a certain male in the city has DNA matching the sample. Suppose, for example, that as part of some study a local hospital had done DNA analyses on patients allergic to something, and a doctor compared the rapist’s DNA with all of the hospital’s DNA records.

Bingo. He finds a match. Nothing connects the man in question with the crime, but the DNA does match. Given that the probability of a match is only one in 100,000, it is almost certain that the man is the criminal — no? No. In fact the probability is very high that the man is innocent. Why? If one in 100,000 men match the DNA from the crime, in a city of 1 million men there will be about 10 men who match the DNA from the criminal.

The fallacy lies in confusing two very different statements. First, and correct: Only one man in 100,000 has DNA matching that of the accused. Second, and false: There is only one chance in 100,000 that the accused is innocent.

This confusion can be dangerous in the hands of unscrupulous or incompetent prosecutors, especially if the jury and defense attorney are unsophisticated. Suppose that (as indeed occurs with serial killers) mutilated bodies are found somewhere. The perpetrator’s DNA is found with the bodies. Two months later a drifter is found three states away with matching DNA. Under interrogation, he confesses to the murders.

The jury understandably is ready to hang somebody, and those numbers — one in 100,000 — sound awfully convincing. The fact is that in a nation of 300 million there will be roughly 3,000 people with DNA matching the killer’s.

DNA evidence is powerful only when combined with other evidence. If the drifter’s DNA matches that at the crime, and his fingerprints are found in a stolen yellow Toyota pickup like one seen in front of the victim’s house, and the footprints in the yard match his Nikes, and witnesses report having seen him in the neighborhood, then he may properly be found guilty.

Other things can render DNA evidence inconclusive, like contamination of the sample. When an impressive technology is presented to people who may not know what it actually represents, it is possible to put someone in jail for something he didn’t do.



Click to Read More

Click to Hide