非対称事前分布を利用した能力母数の推定 : 成長を前提としたテスト場面への適用

概要

論文の詳細を見る
The present study simulated ability estimation based on item response theory, where several tests are conducted during a period in which the academic ability of learners shows improvement, in environments such as language schools. In such a situation, favorable ability estimates can be obtained by developing a prior distribution for the ability of learners based on past measurement results and performing Bayes estimation using the prior distribution. A normal distribution is generally used as a prior distribution; however, in the present study, ability estimation was done using an asymmetric lognormal distribution, which better suits a situation in which ability shows improvement. The results of the present simulation study showed that using a lognormal distribution as a prior distribution, rather than a normal distribution, enabled more favorable ability estimates to be obtained.
文教大学の論文
2004-12-20