Evaluation of a perspective based review method applied in an industrial setting

Inspections are an important means for developing high quality software. The effects of using a perspective based review method with a testing perspective (PBRM(T)) on requirements specifications have been evaluated in an industrial project. PBRM(T) was employed and compared to the normal checklist-based review method (CBRM) which was used earlier in the project. The effects studied were the effectiveness, the efficiency and the number of system faults related to requirements specifications knowledge. Inspection data from CBRM and PBRM(T) were analysed and compared. Problem report data were also analysed and a small survey was used for evaluation of the two methods. PBRM(T) showed an effectiveness of 0.3 faults found per requirement compared with 0.05 with CBRM, or 2.4 faults found per page compared with 0.5 with CBRM in the case study. The efficiency with PBRM(T) was 1.9 faults found per hour compared with 1.2 with CBRM. The engineers expressed that they gained a greater increase in system knowledge when using PBRM(T) compared with CBRM. However, it is not possible in the case study to judge the causality of the effect. The paper indicates that the use of an active review method, such as PBRM(T), yields both a higher effectiveness and efficiency compared with a passive method, such as CBRM, when reviewing requirements specifications.