Hunyh Hunyh
University of South Carolina
BAYESIAN CONSIDERATION IN SOME ASSESSMENT SITUATIONS
Considerations are made on assessment situations that rely on the three- parameter logistic (3PL) and monotone partial credit models. It is proved that, for every prior distribution on ability, modal and mean Bayes
estimates form an increasing sequence in the raw score. For the 3PL model, it is
also proved that the prior that is proportional to the Fisher information
function is the only one that is consistent with the use of the formula for
correcton for random guessing for multiple-choice items. Comments will be made on
how to select items to describe an assessment scale.
Back to Colloquium Series