PubMed1978
Rating scales are used to assess the intensity of psychiatric attributes such as anxiety and depression: they are generally composed of several items to be rated. They postulate: the continuity of the "measured" attribute; a monotonic attribute-item scores relationship; and the cancellation, within the final score of "parasitic" factors due to other features than the one which it is desired to measure. The main metrologic qualities are: sensitivity to states of different intensities, reliability in respect to time and rater, validity: relative to an external clinical classification, to another rating scale, or to known efficacy of a treament on the studied attribute. Self-rating scales avoid an intermediate rater and are easy to use. But they allow for no interpretation from external information, and cannot be used in patients with linguistic difficulties, or severely impaired. Scales rated by a rater consist of codification of a well-conducted interview. They allow for correction of erroneous answers from the patients. They may be affected by certain systematic rater errors (attenuation, central or extreme tendency, halo effect, etc...) which are decreased by training. The global score can be obtained by summing item scores, or from a factor analysis. Construction of a rating scale is a difficult task which implies a series of approximations to obtain, from an initial collection of items, satisfactory metrologic qualities. Translation and adaptation of a rating scale are difficult manoeuvres, which may jeopardize its validity.
Evaluation Studies as Topic↗