Wednesday, March 16, 2011

Should Forensics be Separated from Law Enforcement?

Radley Balko:

After countless scandals in recent years, the problems with America's forensics system are finally getting some national attention. In December, Sen. Patrick Leahy (D-Vt.) introduced a bill to reform the country's crime labs. In January, ProPublica and Frontline teamed up for a year-long investigation into the ways criminal autopsies are conducted across the country. In North Carolina, the state legislature is considering reforms to that state's crime lab, which was rocked by a damning 2010 investigation commissioned by the state attorney general and a follow-up report by the Raleigh News and Observer that uncovered widespread corruption, hiding of exculpatory findings, and a pro-prosecution bias among crime lab workers. All of this comes on the heels of a congressionally commissioned 2009 report from the National Academy of Sciences that found expert witnesses in many areas of forensics routinely give testimony that is not backed by good science.

So the good news is that we are starting to see some skepticism, even some outrage, about the way forensic science is used in criminal cases. The bad news is that the solutions politicians and policy makers are proposing, while better than nothing, do not really address the primary problem. That problem is perverse incentives.

To be sure, there are other problems with the forensics system. For starters, many forensic disciplines, such as hair and carpet-fiber analysis, blood spatter analysis, and especially bite mark analysis, have not been subject to rigorous scientific testing. Even fingerprint analysis is not the sure thing it was once thought to be. Many of these fields were either invented by law enforcement agencies or honed and refined by them. The fields have not been subjected to peer review, and the methods by which, for example, a carpet-fiber or ballistics analyst produces a "match" are not blind. On the contrary, the analyst often knows the details of the crime and which sample implicates the suspect. When done this way, these analyses are not science, but they are often presented in court as if they were.