Abstract:
|
The Center for Epidemiological Studies Depression scale is a multi-item instrument commonly used to measure depression. Item response theory provides a class of models that link observed item scores to an underlying latent trait, such as depression. The traditional use of a total summed score to summarize item score is linked to the partial credit model (PCM) and is shown to be a form of model misspecification in this case. A discrimination-parameter weighted summed score (WSS) is proposed, based on a sample of n = 1278. The WSS is both the minimally sufficient statistic for the estimation of the latent trait and satisfies the monotone likelihood ratio under a Generalized PCM. In patients with chronic illness, such as Systemic Sclerosis (SSc), the measurement of depression is further complicated by disease-related symptoms. Item weighting allows for a measure of depression, and resulting intervention strategies, tailored to this population. The weights show high levels of face validity and reliability and provide meaningfully different ratings of depression. This methodology serves as a guide for the creation of better summary measures for other instruments and other populations.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.