Conceptualizing and characterizing accuracy in assessments of competence to stand trial.

This article describes a mathematical framework for conceptualizing the accuracy of forensic experts' opinions on competence to stand trial (CST) and explains how an expert's expressed opinion about CST can be decomposed into four elements: (1) contextual requirements of the defendant (determined partly by the defendant's past actions) that lie outside the defendant's future control; (2) personal attributes of the defendant that are relevant to competence; (3) the expert's intrinsic ability to distinguish competent from incompetent defendants; and (4) the expert's wish to favor or avoid certain types of outcomes (e.g., a preference to avoid seeing an incompetent defendant stand trial for a serious charge). Because experts are imperfect and have varying levels of confidence in their opinions, one can describe the accuracy of CST assessments by using receiver operating characteristic (ROC) analysis. The article describes some types of insights one might derive from ROC analyses of CST assessments if experts, at least for research purposes, expressed opinions as graded levels of confidence. Although no satisfactory gold standard exists for establishing the truth about a defendant's competence, statistical methods developed over the past two decades may allow investigators to make inferences about the diagnostic accuracy of experts' CST assessments.