The Impact of Missing Data and Multidimensionality on Item Calibration in Computerized Adaptive Testing

• In Event: 29.082 - Roundtable Session 9
  In Roundtable Session: 29.082-12 - Computer Adaptive Testing Designs and Analyses

Abstract

For large operational computerized adaptive testing (CAT) programs, pretest items are often embedded in operational test administrations. Item parameters are then calibrated using the responses collected under this scenario. In practice, this might face two challenges. One is how to deal with the missing data since not all of the pretest items will be administered to a certain examinee. Another challenge is, the assumption of unidimensionality is often violated. In this paper, we investigate the impact of the above two issues by carrying out three simulation studies. The item parameter estimation recovery and examinees’ ability parameter estimation are evaluated under different conditions.

Authors

• Can Shao, National Board of Osteopathic Medical Examiners

• Ying Cheng, University of Notre Dame

• Doyoung Kim, National Council of State Boards of Nursing

• Ada Woo, National Council of State Boards of Nursing