A method for building hybrid defect content and effectiveness models for managing software quality

Quality assurance (QA) plays a crucial role in modern software development. However, methods and models proposed in the literature for supporting quality management suffer from several drawbacks. Many are specialized to certain activities, like system test or inspections. They commonly support only one application purpose, e.g., planning or controlling of QA activities. Often, they are applicable only after measurement data has been collected for numerous historical QA applications or use off-the-shelf models, which do not consider relevant context-specific impact factors. To overcome these drawbacks, we propose a method that can be applied to QA activities during any phase, and which supports comprehensive quality management related tasks: improvement, planning, and controlling. In order to be applicable in practice, the method combines the measurement data typically available with expert judgment to build context-specific models. In addition, the method provides early benefits even if no or few measurement data are available, while motivating the collection of further measurement data by presenting possible improvement directions.