It is often said that replies to a questionnaire may include defective responses that are given by respondents who do not read questions, and that their ratio is higher especially in a web survey. The method to identify defective replies to a questionnaire was proposed and has been evaluated that it can give valid and reliable results. However data analyzed by the method is restricted to semantic differential data which is three mode data. In this paper, we discuss about a method which can extend the coverage of data type to a general two mode questionnaire data.
