[Study of the uni-dimensionality of the Yesavage-Brinck geriatric depression scale. Comparison between classical methods and Rasch's model]

Abstract

The authors present a contribution to the french validation of the self-rating questionnaire of the depression in the elderly proposed by Yesavage and Brink (1982), the Geriatric Depression Scale (30 items). This study focusses on the assessment of the homogeneity and of the unidimensionality of this scale. 99 aged women living in old-people homes or attending a geriatric somatic day-hospital, not known to be psychiatrically ill, filled the GDS and were interviewed by either a psychiatrist or by a clinical psychologist. This interview yielded 44 cases of Major Depressive Disorder or of Dysthymia (DSM III). Firstly, we have applied the classical correlational methods of assessment of scale Reliability and Construct Validity: Cronbach's coefficient alpha and item-total correlations (homogeneity) and Principal Component Analysis (PCA) without rotation. Then, we have performed a Rasch Model Analysis: this method which belongs to the general frame of Latent Trait Theory relies on a probabilistic model of subject's response to individual questions. In the Rasch model, the response probability of a given subject to a given item is a logistic function of the difference between the item location parameter and the subject location parameter along a single continuous latent dimension. Our results have shown that the Cronbach's alpha was very high (.902) and that the item-total correlations were quite satisfactory (mean .470), thus giving a strong impression of homogeneity (similar to unidimensionality for many authors).(ABSTRACT TRUNCATED AT 250 WORDS)