Abstract

Failed software development projects are expensive for society and individual companies. Studies indicate defects in requirements specification are the cause of many of these failures. A large financial services company recently implemented a software requirements inspection process based on the Fagan model with the assistance of the authors. Subsequently, the process was re-invented by users to be less formal, and the organization changed the official process to be consistent with the new process to encourage institutionalization. This change formed a natural experiment. The authors examined inspection documentation under both versions of the process to determine if there were any significant differences in the effectiveness of the versions or their implementation. They found that unplanned implementation effects of the new approach made it impossible to determine whether it was more effective than the original approach. Policy implications are discussed.

Share

COinS