Many existing reputation based trust management frameworks for web services are built on collecting and aggregating the feedback ratings reported by the service consumers. Therefore, the reliability of reputation evaluation mainly depends on the integrity and the accuracy of the reported feedback ratings. In the lectures, various statistical filtering techniques have been proposed to exclude those unfair ratings (a consumer rates a service more positively or more negatively than the real experience). As a kind of unfair ratings, prejudicial feedbacks will obviously reduce the accuracy of the reputation evaluation in the situation that the feedback data are lacking and insufficient. In this paper, we presented our works on rectifying prejudicial feedbacks in a web services management environment. We also presented an experimental study to demonstrate the effectiveness of our rectifying algorithm in increasing the accuracy of reputation evaluation especially when feedback data are not sufficient.