News & Events

Print/PDF this page:

Print Friendly and PDF

Share this page:

Eyewitness Testimony Doesn't Make It True--A Commentary by Steven B. Duke

The following op-ed was originally published in the Hartford Courant on June 11.


Eyewitness Testimony Doesn't Make It True

By Steven B. Duke


The freeing of James Calvin Tillman after 18 years of wrongful imprisonment contains a lesson that has been told and retold thousands of times: Eyewitness identification of strangers is unreliable.

Tillman was convicted of kidnapping and raping a woman in Hartford in 1988 based largely on the testimony of the victim, who had picked out his photo and then identified him in court as the rapist. Eighteen years later, a DNA analysis of semen stains on the victim's clothing showed that they had come from someone other than Tillman. Last Tuesday, Judge Thomas P. Miano set aside Tillman's conviction and set him free.

The mistaken identification of James Calvin Tillman was, unfortunately, a common event and remains so nearly two decades later. Fortunately, in rape and occasional other cases where the perpetrator leaves biological specimens at the crime scene, DNA analysis can today often prevent a miscarriage of justice resulting from misidentification. In many crimes, however, such as robberies and some rapes and murders, there are no physical specimens for DNA analysis. Eyewitness identification remains an invaluable method of convicting those accused of these crimes.

The DNA revolution that began in the late 1980s has dramatically demonstrated how utterly unreliable eyewitness identifications are. About 200 people convicted of violent crimes have been exonerated by DNA evidence in the past two decades. About 80 percent have been the victims of eyewitness misidentification. Some of them served even more time in prison than Tillman.

Even more disturbing are the results of the FBI's DNA analysis of biological specimens in 10,000 cases from 1989 to 1996.

These were all cases in which eyewitnesses had identified a suspect who had been arrested for the crime (usually sexual assault) and biological material from the perpetrator was available for comparison with the suspect's. In 20 percent of the cases, no conclusive results could be obtained. In the remaining 8,000 cases, however, the suspect was cleared in 2,000, or 25 percent. Assuming that without DNA evidence half of these defendants would have been convicted, then as many as 12 percent of those convicted in disputed eyewitness cases may be innocent.

At least 80,000 prosecutions in this country every year rely largely on eyewitness testimony. If only half of those result in convictions, we may still be sending to prison nearly 5,000 innocents annually, based on false eyewitness testimony alone.

DNA has given us a new sense of the dimensions of the problem, but the unreliability of eyewitness testimony is hardly a new discovery. Harvard Professor Hugo Munsterberg made the point in his book "On the Witness Stand" a century ago. Forty years ago, the U.S. Supreme Court opined: "The vagaries of eyewitness identification are well-known; the annals of criminal law are rife with instances of mistaken identification."

Despite our common knowledge that we condemn many innocent people to prison - and sometimes to death - on eyewitness errors, little has been done to reduce the incidence of such errors.

Studies have shown, for example, that if the police who conduct the identification procedures have knowledge of the case and its suspect, they will inevitably influence the eyewitness's memory of the perpetrator in the direction of identifying the suspect. A "double blind" investigation, where both the interrogator and the witness are unaware of the facts of the investigation or the identity of the suspect, is far more likely to produce a reliable identification.

Yet no police department in the country employs this safeguard: The police who investigate the case also interview the witness about the identity of the culprit. If there is a suspect, the interrogator of the witness virtually always knows who that suspect is.

It has also been cogently demonstrated that an identification procedure that exposes photos or people to the witness sequentially, one following another, rather than in a group or lineup is far more likely to result in a correct identification. Yet only a few states employ this method. (A recent study that questions the value of sequential displays is hopelessly flawed.)

Other safeguards are usually rejected. In the 1960s, the Supreme Court ruled that unnecessarily suggestive eyewitness identification procedures are unconstitutional and a witness whose identification is obtained by such procedures is incompetent to testify. In recent years, however, those proclamations have been diluted or disregarded by that court and most others. An eyewitness is virtually everywhere permitted to identify a defendant in court no matter how suggestively her identification testimony was obtained.

Some courts are beginning to acknowledge that special cautionary instructions to the jury are required in eyewitness cases. Yet most still refuse to warn the jury of the fallibility of eyewitness identifications. Some courts allow experts to educate the jury on the nature and sources of eyewitness error, but most do not require trial judges to allow such experts to testify.

Sadly, the window of awareness provided by cases such as James Calvin Tillman's is closing. DNA exonerations are becoming rarer as those mistakenly convicted die or complete their prison sentences, and old case files containing testable material are lost or discarded.

Suspects in new cases where biological evidence exists will usually be tested for DNA, and mistakes like the one that imprisoned Tillman will be reduced. The pressure to reform eyewitness procedures will subside. But we will continue to convict the innocent when DNA is not available to refute the eyewitness.

Once again, as before DNA testing, we will not know which of those we convict are innocent. There will be few exonerations to remind us that thousands are serving time for crimes committed by others who are still free to prey on new victims.

Professor Steven B. Duke teaches criminal procedure and evidence at Yale Law School.