Construct refinement and confirmation for the Addiction Severity Index.

Abstract

A series of analyses reconstruct and confirm the validity and reliability of hypothesized scales for the Addiction Severity Index. Using a diverse sample (N = 990) of methadone maintenance substance abuse patients, a multistage scaling strategy was applied to identify 7 psychometrically integral addiction problem scales. Exploratory item analyses, confirmatory oblique item clustering, and 2nd-order factor analysis verified that the scales comprised relatively little common variance and that each retained a substantial amount of unique and reliable variance. Concurrent and predictive validity were supported with a sample of 244 methadone patients assessed throughout treatment. The advantages of the new scales are discussed, including provision of computer code for calculating normalized standard scores for use in clinical practice and treatment outcome research.