Evaluating distributed inspection through controlled experiments

Inspection methods can be classified according to their discipline and flexibility. The discipline concerns the formal aspect of an inspection method, whereas the flexibility is strongly related to the simplicity of organising and conducting a meeting. The majority of the available distributed inspection methods have a high level of discipline and flexibility as they are based on a well-defined process and the discussion among team members is easily organised and conducted. In this study the authors present two controlled experiments to evaluate the effectiveness and the efficacy of a distributed inspection process to discover defects within source code. In particular, the first experiment compares the distributed inspection method proposed to a disciplined but not flexible method (i.e. the Fagan's inspection process). In the second experiment the authors investigate differences between the same distributed inspection method and a flexible but not disciplined method (i.e. the pair inspection method). Data analysis reveals that more flexible methods require less time to inspect a software artefact, while the discipline level does not affect the inspection quality.