Abstract
STUDY OBJECTIVES: The different versions of the Dysfunctional Beliefs and Attitudes about Sleep Scale (DBAS) have limited comparison and summary of the findings across studies. We aimed to examine which version and which subscales had better psychometric properties. METHODS: Data were derived from a randomized controlled trial of internet-based cognitive-behavioral therapy for insomnia (CBT-I) vs. waitlist in 312 participants with self-report diagnosis of insomnia disorder. The response distribution, internal consistency, construct and concurrent validity, and sensitivity to change were analyzed. RESULTS: Floor or ceiling effects were found in 19 of the 30 DBAS items. Item-total correlation was < 0.30 in 43.3%, 31.3%, and 10.0% of the items in DBAS-30, DBAS-16, DBAS-10. Internal consistency was satisfactory for total scores, with Cronbach α ranging from 0.73-0.81, but 2 subscales of DBAS-30 and 1 subscale of DBAS-10 had Cronbach α < 0.35. Factor analysis produced 8, 4, and 3 factors for DBAS-30, DBAS-16, and DBAS-10. Only the factor structure of DBAS-16 was compatible with previous studies. Concurrent validity with insomnia, anxiety, and depressive symptoms was much stronger than with sleep diary parameters. Sensitivities to change of the DBAS scores following CBT-I and with sleep improvement were found, except the DBAS-30 "attributions" subscale and DBAS-16 "medication" subscale. CONCLUSIONS: The DBAS-16 possesses better internal consistency, a reproducible factor structure, strong concurrent validity, and sensitivity to change, and therefore is recommended for research use. The DBAS-30 and DBAS-10 have their own strengths, but there are limitations in their application as a quantitative measure in research.