Application of the Bifactor Item Response Theory Model for Subscores Estimation

• In Event: Reliability and Subscores

Abstract

One of the widely used latent variables model for large-scale educational testing data is item response theory (IRT). IRT models maximize the information from items in the test and ability of the examinees. Educational testing data typically assesses more than one dominant construct. This paper empirically evaluating subscore estimations by applying unidimensional IRT and bi-factor IRT that embeds the information from general and secondary factors of abilities. Measure of the accuracy and precision of parameters estimation from students’ mathematics abilities with primary and secondary or content domains abilities will be investigated and presented. Finally, subscores from the general and secondary abilities will be estimated and evaluated using the proposed model.

Authors

• Zairul Nor Deana Md Desa, The University of Kansas

• Adibah Abdul Latif, Universiti Teknologi Malaysia