The biasing influence of linguistic variations in DNA profiling evidence

Abstract

Extensive psychological research has confirmed that probabilistic statistical information that is mathematically equivalent, if presented in different linguistic forms, is not psychologically equivalent. Commentators have argued that certain forms of evidence are more prejudicial than probative and should be excluded from trial. In a recent homicide case based on circumstantial evidence, the sole evidence linking a suspect to crime scene was mitochondrial DNA profiling evidence derived from a single loose hair found at the crime scene. Variations in the form of the linguistic evidence presented by DNA forensic scientists and research about its differential impact were cited as the basis for appeal of the conviction to the Australian High Court. A review of the research revealed that presentation of statistical information using natural frequencies is a recommended best practice. Random match probabilities expressed as frequencies are less susceptible of misinterpretation by legal professionals and lay jurors, and lead to fewer false convictions.