Eyewitness Nightmare: We’ve Convicted Countless on Evidence that is Unreliable 25% of the Time

A fundamental principal in American criminal justice is that one is innocent until proven guilty beyond a reasonable doubt. But in the past two decades, DNA-proven wrongful convictions have revealed that we’ve routinely met the standard of “beyond a reasonable doubt” with evidence that is quantifiably incorrect one-fourth of the time.

A 25 percent error rate in school has historically earned the very lackluster grade of D. A 25 percent margin of error would shutter any hospital and ground any airline. But, in the criminal justice system, most Americans, blinded by trust in the system and a popular allegiance to “tough on crime” policies, have yet to demand best practices in securing the most accurate evidence possible from those who have witnessed a crime. This is baffling in light of documented evidence that eyewitness identification is unreliable 25 percent of the time and that best practices can result in more accurate identifications.

DNA-proven wrongful convictions have supported what researchers have known about eyewitness testimony for a century: Our minds don’t work like tape recorders. Even more troubling, memory is subject to contamination that is undetectable to the eyewitness. As Brandon Garrett, University of Virginia Law School professor and author of Convicting the Innocent: Where Criminal Prosecutions Go Wrong, said in a recent Associated Press article on eyewitness policies in lineups, “It’s a very sensitive task. If you do the lineup wrong, you can actually change the memory of the face they saw.”

Presumably police have known this inconvenient fact about this important investigative tool and form of evidence. Studies have revealed that among police lineups in which an eyewitness made a selection, the eyewitness chose a non-suspect filler 30 percent of the time. If a witness fingers a police-selected colleague in the lineup, no harm is done. The police will not arrest him. However, if the witness selects a suspect who happens to be innocent, the stage is set for a wrongful conviction.

This error rate is consistent with the findings of years of pairings in both public and private labs of crime scene biological evidence, primarily from sexual assault cases, with the DNA of the suspect, usually identified by the victim. From a report by Barry Scheck and Peter Neufeld in the landmark 1996 National Institute of Justice study, Convicted by Juries, Exonerated by Science: Case Studies in the Use of DNA Evidence to Establish Innocence After Trial:

“Every year since 1989, in about 25 percent of the sexual assault cases referred to the FBI where results could be obtained (primarily by state and local law enforcement), the primary suspect has been excluded by forensic DNA testing.”

The Innocence Project has consistently reported that mistaken eyewitness identification has been a contributor in about 75 percent of DNA-proven wrongful convictions.

False Justice – Eight Myths that Convict the Innocent, the book former Ohio Attorney General Jim Petro and I co-authored, explores three Ohio wrongful convictions, all resulting from misidentifications. Michael Green spent 13 years in prison before DNA proved he did not rape the critically ill patient who had identified him. Clarence Elkins spent nearly eight years in jail and prison before DNA proved that his six-year-old niece’s statement—that the man who assaulted her in the dark of night looked like Uncle Clarence—was not an accurate identification. The true perpetrator, a pedophile who lived a few doors from the crime, was identified by the DNA testing.

While DNA proved both of these men innocent, it could not save Dean Gillispie. Clothing containing biological evidence in his case was discarded. Gillispie was identified by three victims. Certainly that is foolproof evidence, right? Wrong. In a study of 190 DNA-proven wrongful convictions that included eyewitness misidentifications, multiple witnesses misidentified the same innocent person in 36 percent of the cases, according to Garrett.

I was privileged to write an article on the Gillispie case in the 2011 annual review of the Ohio Innocence Project, which detailed the flawed procedures used in his identification. They violated virtually every best practice in lineup procedures. An article in the Dayton Daily News reported that Gillispie’s photo was all but “starred and circled” in its presentation to the victims. Gillispie never wavered in proclaiming his innocence even when doing so prevented his parole. Gillispie had served twenty years in prison when U.S. District Magistrate Judge Michael Merz granted a Writ of Habeas Corpus and ordered a new trial. Subsequently, the Second District Ohio Court of Appeals vacated Gillispie’s convictions and sentences and also granted him a new trial. He was released from prison in December 2011. The Second District Court of Appeals has upheld its decision in a subsequent motion for reconsideration.

Relying on memory and eyewitness research spanning more than a century and the lessons of DNA, numerous best practices that reduce misidentification in capturing eyewitness evidence have been identified and recommended to police organizations. They have become a familiar litany for those who advocate to police and prosecutor organizations as well as state legislatures in attempts to require them as standard procedure.

These important best practices require a disciplined approach to all procedures in the identification process:

The testimony of the witness should be taken as soon as possible after the crime. Non-suggestive questioning reduces the risk of contaminating the memory. A witness to a lineup should be advised that the suspect may or may not be in the lineup and that the investigation will continue whether or not the witness makes an identification. The filler (non-suspect) persons or photos should be selected to resemble the victim’s earliest description of the perpetrator, as opposed to selections intended to match the suspect. The suspect should not stand out from the fillers, and if the former procedures have this result, the investigation should be broadened. A sequential presentation (persons presented one-at-a-time) requires the witness to make an objective—not a comparative—decision. Blind administration (the administer does not know which person is the suspect) reduces intended and unintended influence. The selection should receive no approval or reinforcement, which increases confidence in the selection, accurate or not.

These recommended procedures cost no more than suggestive procedures, unless you consider the high cost of wrongful conviction; then, of course, the recommended procedures can save taxpayers millions. Nonetheless, these attempts have too often met with resentment and resistance from police and prosecutors, bolstered by arguments that are increasingly indefensible.

Recently, a bill that would have required best practices in eyewitness procedures was passed in the Florida Senate but died after not making it to a full House vote. Scott Maxwell’s article in the Orlando Sentinel properly chastised the state’s political leaders and voters for not recognizing the importance of addressing the state’s wrongful conviction problem. Florida leads the nation in former death row inmates now exonerated (23).

Unfortunately, Florida is not alone in resisting requiring improved procedures. Even as Americans have seen increasing evidence of the frightening scope of wrongful conviction, preventable crimes perpetrated by the real criminals who escaped justice, a parade of ruined lives of the wrongfully convicted, and millions wasted in misspent taxpayer dollars, most voters remain complacent or even supportive of failed practices in criminal justice.

We’ve had too many wakeup calls. Responsible citizens must urge our elected officials to require best practices in the criminal justice system.

Any case based solely upon eyewitness identification should raise red flags. It’s becoming increasingly obvious that eyewitness identification alone, with its 25-percent error rate, cannot prove guilt beyond a reasonable doubt.