Reliability and validity of drug use items differing in the nature of their response options.

A consensus among researchers about the best way to measure self-reported drug use has not yet emerged (Huba & Bentler, 1982; Huba, Bentler, & Newcomb, 1981). The average product-moment correlation between 14 pairs of experts who rated the adequacy of 17 drug-use quantity and frequency items was only 0.12 (Huba, Bentler, & Newcomb, 1981). Criteria used to evaluate drug-use measures are implicit, and item selection tends to be guided by personal preferences and intuition. Response formats used in drug-use surveys have varied from qualitative (such as Huba, Wingard, & Bentler, 1981) to quantitative (such as Huba, Bentler, & Newcomb, 1981) to extremely quantitative (such as the open-ended items of Jessor & Jessor, 1977). The implications of using different response options are unclear and have been the subject of recent debate (Baumrind, 1983; Huba & Bentler, 1982; Martin, 1982). In the present study, we compared items with response formats that have been commonly used to measure drug use. Items were chosen to provide a representative sample of response options that differ in the degree of their quantitativeness. We evaluate these items using standard psychometric criteria.