An Application of Item Response Theory to the Measurement of Depression

Studies of group differences in a subjective phenomenon are generally concerned with dzfferences in the degree of the trait. To determine differences in degree, however, it is first necessary to determine that the items used to measure the trait in the groups do not differ in kind. In other words, the items must be unbiased. Evaluating differences in kind and in degree requires the ability to scale questions and respondents. Item response theory (IRT) models offer a method for scaling questions and respondents with respect to a subjective phenomenon. These models state that the probability of a correct response depends on the level of the trait the person has and on the characteristics of the item being answered. This paper presents an application of the two-parameter logistic IRT model to the measurement of depression.