RELIABILITY ENHANCEMENT AND ESTIMATION IN MULTISITE CLINICAL-TRIALS

Clinical trials, particularly those in addictions research, often rely on self-report data for primary dependent variables, and it is impera tive to evaluate and to minimize both random and systematic error. Thi s article describes methods for enhancing and assessing reliability of measurement in multisite clinical research. It begins with an overvie w of the two major approaches to evaluating the reliability of psychia tric diagnosis and verbal self-reports. A model of the interview proce ss is then described, and the major sources of inconsistency that aris e in the data collection process are identified Based on the model, st aff selection criteria training techniques and quality assurance proce dures are recommended for enhancing the reliability of interview asses sments, and a research design appropriate for evaluating reliability i n multisite clinical investigations is proposed.