An important observation in these data was that (a) nearly all those who endorsed craving met criteria for DSM-IV alcohol dependence and (b) craving was more commonly reported in those with a greater number of alcohol dependence symptoms, which is consistent with a prior latent class analysis in an extended subset of these data (Foroud et al., 1998). These findings, despite being drawn from a study ascertained for alcoholism, are consistent with the extant epidemiological literature that notes that craving, while being an excellent item when distinguishing between those at high versus lower liability to alcohol dependence, is only effective at making this distinction in individuals at high risk. In the context of psychometrics, the craving item has high discrimination (i.e. good factor loading) and high difficulty/severity (i.e. infrequently endorsed by low risk individuals). For instance, in the National Longitudinal Alcohol Epidemiologic Survey (NLAES), craving cohered well with other DSM-IV alcohol dependence criteria, had the second highest discrimination and also the highest severity (with the exception of the abuse criterion of legal problems, which is proposed for exclusion in DSM-5).
