In psychometric analyses, craving is considered a “difficult” item – that is, it is infrequently endorsed and even though it discriminates between individuals with and without substance use disorders, it does so primarily in individuals already at high liability to develop the disorder. In other words, those who report craving are typically already classified as substance dependent due to endorsement of multiple (often more than 3) dependence criteria; craving further discriminates these most severely affected individuals (i.e. those with multiple other symptoms) (Bohn, Krahn, & Staehler, 1995; Bucholz et al., 1996). Consequently, the addition of craving as a criterion does not produce a significant increase in those diagnosed with alcohol use disorders (Mewton et al., 2010; Agrawal, Heath, & Lynskey, 2011; Keyes et al., 2011; Cherpitel et al., 2010). This opens the door to the discussion of whether craving is best assessed using a single item or whether the architecture of craving is better suited to a multi-dimensional assessment in order to provide unique information beyond the existing definitions of dependence.
