A0371
Title: Optimizing test item calibration - with application to the Swedish national mathematics test
Authors: Ellinor Fackle-Fornius - Department of Statistics (Sweden) [presenting]
Frank Miller - Stockholm University (Sweden)
Abstract: For large-scale achievement tests, like national tests in school, test items need to be calibrated before being implemented in the test. Item calibration is the process of estimating item parameters such as difficulty and discrimination through pretesting. It's crucial to estimate the item parameters with as high precision as possible, as it influences the test quality and the accuracy of ability estimates for the examinees. Instead of randomly allocating calibration items to examinees, optimal design theory is utilized to allocate items to examinees in an optimal way based on their individual ability levels. An ability-matched item allocation method is employed, tailored to handle item calibration conducted in large groups in parallel with items of mixed formats, a common scenario for many large achievement tests. The optimal design, IRT analyses, and results of a real calibration study conducted for the national test in mathematics in Sweden are presented.
