Parameters and Models of Item Response Theory (IRT): A Review of Literature

Abstract

Introduction: Item response theory (IRT) has received much attention in validation of assessment instrument because it allows the estimation of students’ ability from any set of the items. Item response theory allows the difficulty and discrimination levels of each item on the test to be estimated. In the framework of IRT, item characteristics are independent of the sample and latent traits of the person are independent of the test on the account that the selected models perfectly fit the data. Therefore, scores that describe examinee performance are independent on test difficulty. The scores of the examinee may be lower on a difficult test and higher on easier tests, but the ability level of the examinee remains the same over any test at the time of testing. The IRT model allows the estimation of item parameters. The line of difference between the models and parameters of IRT is not clear to many students in assessment. Purpose: This paper reviews the parameters that are estimated using IRT and the models available in IRT. Also, the paper highlights the difference between the parameters and models and the various models under each set of data. Methods: Various literatures on IRT relating to the parameters and models of IRT are reviewed. Conclusions: There are four parameters estimated with IRT but the models are not four. Again, the models of IRT depends on the type of data. Dichotomous data has four models for the four parameters. However, polytomous data has two parameters: item difficulty and item discrimination for the models.