Bayesian Estimation of Test Engagement Behavior Models with Response Times
Kazuhiro Yamaguchi, Kazuya Fujita · 2022
Detecting non-engagement test answering behavior is a crucial task in situations where the tests are low-stakes for the individuals but the scores are employed in decision-making. Nagy and Ulitzsch (2022) proposed four test engagement models, but their estimations were conducted using maximum likelihood estimation. This study applied the Bayesian estimation method to the test engagement models. Bayesian formulation of the test engagement models was introduced and estimation was conducted using “just another Gibbs sampler” language. Real data analysis was conducted and problems were discussed regarding the future orientation of the Bayesian framework in test answering behavior modeling.