Psychology Commons^™

A Positive Trait Item Response Model, Joseph F. Lucke

Joseph Lucke

All current models from item response theory (IRT) assume the latent trait follows a standard normal distribution. While this assumption is appropriate for traits such as ability or attitude, it creates both conceptual and technical problems traits such as addiction (alcohol, drugs, gambling). The distribution of an addiction trait is better assumed to be anchored at zero (no addiction) and positively skewed. A small change to the usual IRT model yields a class of positive-trait item response models (PTIRMs). I discuss PTIRMs and present one model in detail, including item characteristic curves and item information curves. I present an example …

An Introduction To Item Response Theory For Health Behavior Researchers, Russell Warne

Russell T Warne

OBJECTIVE:

To introduce item response theory (IRT) to health behavior researchers by contrasting it with classical test theory and providing an example of IRT in health behavior.

METHOD:

Demonstrate IRT by fitting the 2PL model to substance-use survey data from the Adolescent Health Risk Behavior questionnaire (n=1343 adolescents).

RESULTS:

An IRT 2PL model can produce viable substance use scores that differentiate different levels of substance use, resulting in improved precision and specificity at the respondent level.

CONCLUSION:

IRT is a viable option for health researchers who want to produce high-quality scores for unidimensional constructs. The results from our example-although not …

Positive Trait Item Response Models, Joseph F. Lucke

Joseph Lucke

A new item response model is proposed for which the trait is positive. Three such models, the loglogistic, the log-normal, and the Weibull, are presented along with their item information curves. The data of seven addiction items from the DSM-IV from a study on alcohol addiction is analyzed by these three models using Bayesian Markov chain Monte Carlo methods. The item characteristic curves and item information curves are presented for all three models. The person scores for four item response patterns are presented for the log-logistic model.