抄録
A questionnaire survey on indirect speech requests conducted by Hirakawa was analyzed using Item Response Theory (IRT) applying Bayesian statistical modeling. From the knowledge of both the characteristics of the respondents and the characteristics of the question items, it is found that the characteristics of the question items are the cause of the bimodal distribution of respondents tendency to interpret the interpretation of the request. Also Bayesian IRT appears to offer benefits for scale design.