Source Code Inspection: Measuring the Difference Between Individual and Group Performance

Abstract

The purpose of this study is to examine the relationship between the performances of group efforts versus individual efforts. Study of previous research on software inspection has shown that opinions are divided on whether to have a group work stage in the inspection or not. Initially believed to be beneficial for finding defects in source code it has in more recent years shown to be not as effective as once believed and possibly not worth the cost. This study examines the result of an experiment that conducts a slightly altered version of software inspection where subjects work either solely in groups or solely as individuals where the traditional method has subjects first inspect the code alone and then forming a group to discuss their findings. The hypothesis was that the subjects working in groups in such a fashion will perform better, but it is shown that neither group of subjects outperforms the other at finding defects. These results are consistent with previous research in the field, however it is interesting to note that even without individual preparation groups perform similarly to how they have performed in other studies. Since one of the chief arguments against group work is its high cost eliminating a stage from the review process would make it less costly and thus perhaps more desirable as an inspection method.