Abstract This special section introduction contrasts early skepticism in forensic science about using proficiency tests to estimate examiner error rates with contemporary support for deriving error rates from black box studies. It highlights a central methodological dispute: whether and how “inconclusive” conclusions should be incorporated into error-rate calculations. Luby and Kadane propose a variance-decomposition framework that reframes the debate by distinguishing examiner-driven from item-driven sources of inconclusives, and introducing “failure rates” as an alternative metric. The three commentaries and rejoinder that follow advance the discussion about how forensic inconclusives should be analyzed and communicated.
Jonathan J. Koehler (Thu,) studied this question.