Enhance Classification Accuracy in Computerized Adaptive Testing via Test Administration Algorism

• In Event: 72.066 - Poster Session 16
  In Poster Session: 72.066-2 - Poster Session: Practical Issues in Measurement

Abstract

For most assessment programs using CAT, it is important to provide accurate ability estimates and also correct classification decisions. Classification errors happen most often to the “threshold” examinees – whose abilities are close to the cut scores. This study aims to illustrate the possible improvement in classification accuracy by varying the exposure rate around the cut scores. Various conditions were simulated with fixed and variable length CAT. Preliminary results suggest that the administration of the item of best information around the cut scores led to somewhat increased classification accuracy. Treating the uniform group size of 1 as a baseline, varying item exposure around the cut scores resulted in comparable classification accuracy while keeping test overlap rate lower across ability ranges.

Authors

• Jie Lin, Pearson

• Che-Ming Allen Lau, Pearson