Journal of Forensic Research

When Good DNA Goes Bad

Abstract

Mark W. Perlin

DNA evidence is the forensic gold standard. However, the interpretation of this evidence can be challenging. Sophisticated mathematical computing can provide accurate and reliable interpretation of DNA mixtures that contain two or more individuals. But the reliability of human review of such data is less well established. This paper explores what happens when good DNA data is badly interpreted. The genotype is used as a unifying concept, with mixture evidence containing multiple genotypes. Highly informative computer interpretation of mixtures is compared with less quantitative human approximation. Seven cases are examined where, on the same data, human review gave appreciably different results than the computer’s more thorough assessment. These interpretation differences can affect criminal justice outcomes, suggesting that sophisticated computing is needed to help people interpret challenging DNA mixture evidence.