Table of Contents | Search Technical Documentation | References
Estimation of IRT Item Parameters
Treatment of Missing Responses in NAEP
Assumptions of the Item Scaling Models
Checks for Violations of Assumptions
Treatment of Items to Avoid Violations of Assumptions
NAEP score scales summarize statistics describing scale scores for groups of students for the collection of items representing the content specified in the NAEP frameworks. For each subject area, the framework determines the number of Item Response Theory (IRT) scales.
IRT models are used to describe the relationships between the item responses provided by students and the underlying score scales. IRT provides a common scale on which the performance of students receiving different blocks of items can be placed. For each item, item parameters that are used in the models are estimated from student response data. Different IRT models with different types of item parameters are used to describe multiple-choice items, constructed-response items that are scored right or wrong, and constructed-response items that have more than two categories. However, all types of items contribute to NAEP score scales. Evaluation of model fit is done during the item parameter estimation procedure.