For the evaluation of psychometric properties of screening instruments it is important to differentiate between studies using samples from general populations and those using specific user populations, as this directly affects several psychometric indices. For example, the prevalence of a disorder within the study population will influence the ability of any screening test to detect it accurately. Thus, if cannabis dependence is rare (in general population samples), a negative test will more likely indicate no disease (high NPV) and a positive test will less likely indicate the presence of the disease (low PPV) [27] than when it is common, e.g. in samples of heavy cannabis users.
