A Bayesian Logistic Regression Method to Calibrate Pretest Items in Multistage Testing

• In Event: Electronic Board Session I

Abstract

The Bayesian logistic regression (BLR: Ho, 2023) method was developed to address the response sparseness of linking items leading to concern about the quality of the link to the item bank in multistage testing. An optimized approach is proposed to improve the BLR performance evaluated using the simulation data.

Author

• TsungHan Ho, ETS