Mixed Adaptive Multistage Testing: A New Approach

Authors : Anthony RABORN, Halil SARI

Pages : 358-373

Doi:10.21031/epod.871014

View : 69 | Download : 17

Publication Date : 2021-12-29

Article Type : Research Paper

Abstract :Computerized adaptive testing (CAT) and computerized multistage testing (CMT) are two popular versions of adaptive testing with their own strengths and weaknesses. This study proposes and investigates a combination of the two procedures designed to capture these strengths while minimizing the weaknesses by replacing the standard MST routing module with a CAT-based, item-level routing module. A total of 3000 examinees were simulated from a truncated normal distribution with bounds at -3 and 3, and a simulation study was conducted. Simulation results indicate that the new method provides some efficiency improvements over traditional MST when both routing modules are the same size, and when the item-level routing module is larger, the improvements are greater. The study showed that the proposed test administration model could be used to measure student ability, meaning that our new method resulted in lower mean bias, lower RMSE, and higher correlation than traditional MST. An R package built from the code used for this paper is also introduced in the supplementary file. The limitations of the study and recommendations for future research are also presented.
Keywords : Computerized adaptive test, multistage adaptive test, simulation, R, mixed adaptive test

ORIGINAL ARTICLE URL