New Efficient and Practicable Adaptive Designs for Calibrating ItemsOnline

摘要

When calibrating new items online, it is practicable to first compare all newitems according to some criterion and then assign the most suitable one to thecurrent examinee who reaches a seeding location. The modified D-optimal designproposed by van der Linden and Ren (denoted as D-VR design) works within thispracticable framework with the aim of directly optimizing the estimation of itemparameters. However, the optimal design point for a given new item should beobtained by comparing all examinees in a static examinee pool. Thus, D-VR designstill has room for improvement in calibration efficiency from the view oftraditional optimal design. To this end, this article incorporates the idea oftraditional optimal design into D-VR design and proposes a new onlinecalibration design criterion, namely, excellence degree (ED) criterion. Fourdifferent schemes are developed to measure the information provided by thecurrent examinee when implementing this new criterion, and four new ED designsequipped with them are put forward accordingly. Simulation studies wereconducted under a variety of conditions to compare the D-VR design and the fourproposed ED designs in terms of calibration efficiency. Results showed that thefour ED designs outperformed D-VR design in almost all simulationconditions.