The Arizona Republic

Phoenix police lab errs on DNA

9 cases under review after mistakes found

Carlos Miller
The Arizona Republic
May. 6, 2003 

Phoenix police crime lab technicians blundered nine cases while analyzing DNA evidence to be used against murder, rape and aggravated assault suspects.

The errors, which dated to August 2001, were made when the technicians miscalculated the likelihood that a person's DNA, or genetic material, was present on evidence.

Although the discovery leaves the fate of the nine suspects in limbo, police said they have plenty of other evidence, including witness statements, confessions, gunshot residue and other DNA samples.

"I don't expect any of these people to get out of jail or prison," Phoenix police Sgt. Randy Force said.

Maricopa County Attorney Rick Romley, who was informed of the situation last week, is reviewing the cases. Defense attorneys, meanwhile, are informing their clients of the lab mistakes.

Two of the suspects have pleaded guilty and were sent to prison. A third suspect was sent to prison after a jury found him guilty. Charges were never filed against a fourth suspect and five other suspects are awaiting trial.

Police said insufficiently trained lab technicians using a rare method of DNA testing were to blame for the "statistical computation errors." The method, called the "likelihood ratio system," involves the use of swabs on pieces of evidence. It is used in fewer than 10 percent of the department's cases involving DNA sampling, Force said.

Some agencies prefer not to use it because it can be difficult to explain to a jury in layman's terms, he said.

Unlike the standard methods of DNA sampling, in which lab technicians analyze samples of blood or semen that originate from a single source, the likelihood ratio system analyzes objects that have been handled by several people.

As a result, lab technicians may end up analyzing DNA from several people before determining whether the suspect can be linked to the crime.

Once a suspect's DNA is isolated from a piece of evidence, technicians use a mathematical computation to determine the likelihood that a suspect committed a crime by using five categories, ranging from "limited evidence to support" the case to "very strong evidence to support" it.

This is when technicians erred.

In one of the nine cases, in which a murder suspect is awaiting trial, evidence against him went from "very strong" to "limited." Evidence against another man, who is also awaiting trial on murder charges, went from "very strong" to "moderate." The names of the suspects were not released.

The blunders came to light last week after a lab supervisor discovered an error in one of the cases, prompting officials to go back three years to when the lab began analyzing DNA samples. It reviewed about 40 cases that entailed the likelihood ratio system.

"It's an honest mistake made with the best intentions," said Susan Narveson, administrator of the department's crime lab.

Police said they have taken measures to ensure this will not happen again, including inviting outside auditors to review lab operations.

"The goal of the crime lab is to guarantee 100 percent accuracy, 100 percent of the time," Force said. "We have fallen short of that goal."

Reach the reporter at or (602) 444-8208.

Junk Science
Truth in Justice