People v. Nelson: A Tale of Two Statistics

Abstract

In recent years, defendants who were identified as a result of a search through a database of DNA profiles have argued that the probability that a randomly selected person would match a crime-scene stain overstates the probative value of the match. The statistical literature is divided, with most statisticians who have written on the subject rejecting this claim. In People v. Nelson, the Supreme Court of California held that when the random-match probability is so small as to make it exceedingly unlikely that any unrelated individual has the incriminating DNA profile, this statistic is admissible in a database-search case. In dicta, the court suggested that the defendant might be permitted to introduce an inflated match probability to counter the prosecution's statistic. This Comment describes the statistical issue, questions some of the reasoning in Nelson, and suggests other approaches that a defendant might take in response to a cold hit in the database.

Keywords: evidence, probative value, likelihood, probability, random match, database trawl, DNA profiling, np rule, Bayes' rule