Integration of Analytical Quality Assurance Methods into Agile Software Construction Practice Research Proposal for a Family of Controlled Experiments

Defects in early software development products, e.g., design specifications, can have a major impact on product quality. Analytical quality assurance methods, like Software Inspections and Testing are common practices for detecting non-conformities in software products. Often, these quality assurance (QA) activities are not seen as integral part of software construction process but rather as time-and-budget-consuming add-on activities, which may be cut short if the remaining budget becomes tight. Agile software development practices, such as pair programming, focus on software construction, with some basic elements of quality assurance, e.g., continuous reviews. Several empirical studies reported on defect reduction in software products and on productivity improvements of pairs compared to solo programming. To bundle the benefits from agile development practice and structured quality assurance activities, we propose a pair programming (PP) approach that integrates best-practice inspection and usage-based testing (see ISERN 2004 poster session). This approach is expected to bring several benefits: (a) increased product quality, (b) increased business value contribution of PP by applying prioritized use cases, (c) improved understanding of customer requirements, and (d) extending application opportunities of agile methods to more critical software systems, which require auditable QA activities. Based on this integrated PP model we propose to conduct a family of experiments to investigate defect detection capability and learning effects. Additionally, the experiment setup will address the theoretical background (Empirical Software Engineering theory building) on the necessary number of participants to achieve empirical evidence, improve reporting theory of empirical studies and improve hypothesis building processes.