Improving the Measurement of Psychopathology

University Of North Carolina Chapel Hill, Chapel Hill NC

Investigators

Linked publications & trials

Abstract

DESCRIPTION (provided by applicant): The purpose of this research is to improve measurement of psychopathology through the use of item response theory (IRT). IRT is a set of statistical models and techniques used to evaluate items and score questionnaires, interviews, rating scales, etc. IRT techniques are preferable to classical test theory methods (e.g., Cronbach's alpha, summed scores), because IRT models provide detailed information about items that: a) aids in the creation of short, but reliable instruments, b) produces indices of scale precision that vary for different people to better reflect individual differences, and c) allows for the computation of scores that account for heterogeneity of discrimination ability among the items. However, currently implemented IRT models often assume that the construct measured by the items is normally distributed in the population, which is untenable for psychopathology. IRT analyses based on a false assumption about the distribution of the latent construct likely produce misleading results. The first aim of the proposed study is to adapt existing IRT so that the distribution of a latent construct in the population is allowed to adopt nearly any shape. The new procedure and software will be tested with a computer simulation study, wherein the algorithm is applied to computer-generated data. The second aim of the proposed study is to apply the new technique to data from two self-report questionnaires on obsessive-compulsive disorder symptoms, thereby producing distributions of the latent constructs and properties of the items, which may suggest potential revisions to the measures.