Item response theory for measurement validity

Item response theory (IRT) is an important method of assessing the validity of measurement scales that is underutilized in the field of psychiatry. IRT describes the relationship between a latent trait (e.g., the construct that the scale proposes to assess), the properties of the items in the scale, and respondents’ answers to the individual items. This paper introduces the basic premise, assumptions, and methods of IRT. To help explain these concepts we generate a hypothetical scale using three items from a modified, binary (yes/no) response version of the Center for Epidemiological Studies-Depression scale that was administered to 19,399 respondents. We first conducted a factor analysis to confirm the unidimensionality of the three items and then proceeded with Mplus software to construct the 2-Parameter Logic (2-PL) IRT model of the data, a method which allows for estimates of both item discrimination and item difficulty. The utility of this information both for clinical purposes and for scale construction purposes is discussed.