Reliability and Validity

Background: To assess outcomes of health services, providers need brief, responsive, reliable, and valid measures that can be implemented in clinical settings with minimal cost and burden. The Behavior and Symptom Identification Scale (BASIS-32) is a selfreport measure developed in 1984 to assess mental health treatment outcomes. During the past 3 years, multiple methods were used to revise the instrument to improve reliability, validity, and applicability to diverse groups of mental health service recipients. Objective: The objective of this study was to field test the revised instrument, make further changes based on analysis of the field test data, and assess reliability and validity of the final version (BASIS24). Methods: A field test was implemented at 27 treatment sites across the Untied States. A total of 2656 inpatients and 3222 outpatients participated. Factor analytic methods, classic test theory, and item response theory modeling were used to select the most discriminating, nonredundant items for inclusion in the final version of the instrument and to assess its reliability and validity. Item response theory modeling was used to score the instrument. Results: The final instrument includes 24 items assessing 6 domains: depression/ functioning, interpersonal relationships, selfharm, emotional lability, psychosis, and substance abuse. Test–retest and internal consistency reliability were acceptable. Tests of construct and discriminant validity supported the instrument’s ability to discriminate groups expected to differ in mental health status, and its correlation with other measures of mental health. Conclusions: Analyses of the BASIS-24 supported its reliability and validity for assessing mental health status from the patient’s perspective.