Forensic experts have been retained by both sides in criminal and civil court proceedings in the United States for decades. In 1993, the Daubert decision by the U.S. Supreme Court and decisions that followed established some criteria to determine what constitutes “good science” and reliable testimony. Several of the Daubert criteria explicitly refer to statistical ideas, including uncertainty and error, validity, and reproducibility.
In this presentation, I revisit the Daubert decision wearing a statistician’s hat. Using examples from various forensic disciplines, I examine where they may fall short relative to the Daubert criteria and discuss some of the ongoing work in statistics that will help fill the gap.
Statisticians who work in forensic statistics can have a tremendous impact on the fair administration of justice. The Innocence Project estimates that about 20,000 incarcerated persons are not guilty of the crime of which they were convicted. Two contributing reasons for the high number of wrongfully convicted persons are incorrect forensic analyses or conclusions that exceed what the results of the analyses can support.
|