Defect content estimations from review data

Reviews are essential for defect detection and they provide an opportunity to control the software development process. This paper focuses upon methods for estimating the defect content after a review and hence to provide support for process control. Two new estimation methods are introduced as the assumptions of the existing statistical methods are not fulfilled. The new methods are compared with a maximum-likelihood approach. Data from several reviews are used to evaluate the different methods. It is concluded that the new estimation methods provide new opportunities to estimate the defect content.