Estimation of Item Parameters as a Method for Supporting Learner/Examinee Assessment

Item Response Theory (IRT) is a model for expressing the association between an individual's response to an item and the underlying latent variable (ability) being measured by the instrument. Item Characteristic Curves (ICCs) are one of the basic blocks of an Item Response Theory, nd their parameters (difficulty, discrimination and guessing) must be estimated accurately. The estimated parameters will subsequently be used to form ICCs of an exam upon which other latter judgments about examinees’ abilities will be made. Regarding the importance of assessment in learning process and reaching accurate estimations about learners' abilities, this paper is focused on a comparative approach for finding the best technique of estimating these parameters. The criterion for such an optimization is the chi-square goodness of fit. Results show that Genetic Algorithms obtain the best estimations among two other applied techniques.

Estimation of Item Parameters as a Method for Supporting Learner/Examinee Assessment Keywords:

Item Response Theory , Item Characteristic Curve , Intelligent Tutoring System , Assessment , Genetic Algorithm , Simulated Annealing , Gradient Descent