May 30, 2002

On the Evidence

Juries in criminal trials are asked to decide if the evidence proves the defendant's guilt beyond a reasonable doubt. What "reasonable" means always has been open to question, and the question has grown murkier in recent years. Evidence once considered reliable has fallen under strict scrutiny, with results that are less than heartening.

Eyewitness testimony is among the shakiest. Numerous experiments have shown how unreliable witnesses can be. In one study, half the participants asked to watch a tape of a basketball game and count the passes failed to notice a woman in a gorilla suit appear on screen and beat her chest vigorously. Cases that rely on witness testimony for a conviction are not always necessarily the strongest, and Times-Dispatch reporter Frank Green recently has detailed some of those in Virginia open to question.

Another form of evidence, handwriting analysis, also has been challenged. Daubert v. Merrell Dow Pharmaceuticals - a 1993 Supreme Court case dealing with what has come to be known as "junk science" - gave judges discretion over whether to accept expert testimony, regardless of their own expertise in a given field. In two recent cases, judges came to opposite conclusions about the validity of handwriting analysis, with one deeming it too unscientific to be admissible.

The court that upheld handwriting analysis relied on a new study to appear in The Journal of Forensic Sciences that offers what its author claims is the most comprehensive proof that each person's handwriting is unique. Special computer programs were able to identify writers with a 98-percent level of confidence. Human analysts, according to the study, could then examine the "finer features" of a handwriting sample and raise the accuracy of identification to nearly 100 percent.

Now there is word of doubt concerning the reliability of fingerprints as evidence. No two individuals have the same set of prints. But fingerprint identification does not necessarily offer the degree of precision seen in DNA identification. Prints at crime scenes can be messy - half a print in a smear of blood, or a partial print on an uneven surface, for instance, which must be examined for comparison points with the neatly rolled prints made at police stations.

The FBI's Integrated Automated Fingerprint Identification System in West Virginia holds a vast array of computers that do the initial check on the 40,000 sets of prints received each day. The computers make an accurate identification 99.97 percent of the time. That sounds impressive, unless you are one of the 300 mistakes in each million searches. When the computers cannot tell the difference between similar sets of prints, human analysts - and therefore subjectivity - take over. Yet the element of subjectivity undercuts the claim to scientific objectivity a computer can support. Both a New York Times story on handwriting analysis and a New Yorker story on fingerprinting contain quotes to the effect that belief in the evidence ultimately must rely on faith.

But then, so must the entire judicial system. Judges are human. So are prosecutors, defense attorneys, police officers, and jurors. As the use of DNA testing has shown, forensic evidence can build a powerful, sometimes almost irrefutable, case. Yet even if forensic tests should one day attain 100-percent accuracy, the public must have faith that the evidence was not planted, tampered with, switched, suppressed, or otherwise manipulated; that confessions have not been coerced; that a deal has not been cut; that the defense attorney is not rushing through a pro bono case to get back to one that generates fees; and so on.

In short, the public is learning that there is room for improvement in the treatment of evidence - but that no amount of improvement can obviate the need for faith in one's fellow citizens. In the pursuit of justice, that is an ancient - and seemingly eternal - truth.

Junk Science
Truth in Justice