Inconclusives are not errors: a rejoinder to Dror

In Arkes and Koehler (2022), we proposed that signal detection theory provides a helpful framework for understanding when and why a forensic examiner may wish to call an ‘inconclusive’ rather than offer a binary source conclusion (i.e. exclusion or identification) when deciding whether two samples share a common source. As Dr. Itiel Dror notes in his thoughtful letter to the editor, inconclusive decisions by forensic examiners have important implications for both casework and error rate studies. If there is a higher rate of inconclusive decisions in error rate studies than in casework, then the studies probably underestimate the casework error rate when inconclusives are set aside. We agree with Dr. Dror on this point. We part company with Dr. Dror when he proposes that inconclusives should be scored as errors in test situations ‘when there is enough information to determine the state of nature’ (Dror, in press; see also Dror and Scurich, 2020). We disagree with this proposal for two reasons.