Comparing Item Selection Methods in Multidimensional Computerized Adaptive Testing Under Different Types of Multidimensional Structures

• In Event: 65.079 - Roundtable Session 23
  In Roundtable Session: 65.079-14 - Issues in Test Assembly

Abstract

The performance of item selection methods in multidimensional computer adaptive testing (MCAT) have always been studied under simple multidimensional structures. The current study aims to explore the performance of different MCAT item selection methods under two different multidimensional factor structures: simple vs. complex. For each type of structure, two, three and four dimensions are considered. Five item selection methods are Kullback-Leibler information-KL, Minimum Angle-Ag, Volume-Vm, Minimize the error variance of the linear combination-V1, and Minimize the error variance of the composite score with the optimized weight-V2. Two different termination rules are used. The recovery of domain scores and overall scores, measurement precision, and absolute bias are used as evaluation criteria. Other conditions are fixed for all other conditions.

Authors

• Halil Ibrahim Sari, University of Florida

• Cengiz Zopluoglu, University of Miami

• Anne Corinne Huggins-Manley, University of Florida