Psychometric considerations of depression symptom rating scalesAuthor(s): Shawn M McClintock, Charlotte Haley, Ira H Bernstein
A major difficulty associated with the diagnosis of major depressive disorder (MDD) is that it cannot be unequivocally tied to observable bodily functions in the sense that a condition like hypertension can be described in terms of elevated blood pressure. Thus, MDD exemplifies what is called a ‘latent variable’, ‘inferred variable’ or ‘construct’. Creation of scales to measure MDD symptomatology in adult populations is a popular line of research. The traditional approaches to constructing and evaluating scales are known as classical test theory, but modern test theory – more specifically, item response theory – offers some advantages. One useful feature is the ability to equate a score of X on scale A with a score of Y on scale B. Both classical test theory and item response theory are examined in this review. Perhaps the main point is that scales for depression, whether based upon self reporting or clinician rating, use fairly similar item content so that they are more similar than different. Thus, the purpose of this paper is to review the general considerations involved in constructing depression scales, which applies to the many other latent variables studied in psychiatry, and to compare important factors regarding the utility of scales for research and clinical practice.