Arne Beckhaus

Modelling software quality costs by adapting established methodologies of mature industries

In the software sector, the calculation and modelling of quality costs have not achieved the maturity of other industries yet. In this paper, a model that bridges the gap between classic cost accounting and software development specific models is proposed. This model considers defect introduction and removal, and can be applied to different granularities. An implementation plan is suggested for its instantiation and explained exemplarily by a commonly used software quality model.

The Impact of Collaboration Network Structure on Issue Tracking's Process Efficiency at a Large Business Software Vendor

Researchers in the IS domain have addressed communication structure and its effect on performance. While early research focused on small networks and utilized sociometric surveys, recent works have concentrated on electronic data sources provided by open source software repositories. Surprisingly, software vendors have less frequently been studied despite their need for continuous enhancement of organizational design. In this study, we analyze a global business software vendor by utilizing existing data sources. We investigate the association of both collaboration network centrality and communication with process efficiency. Despite coping with complex, interweaved processes and social networks, we find that communication, centrality in the large case-study-wide network, and communication pattern homogeneity are positively associated with process efficiency. However, centrality in small work groups slows the analyzed issue tracking process down, possibly due to increased overhead and bottleneck effects that become visible when looking at single issues qualitatively.

Team Factors and Failure Processing Efficiency: An Exploratory Study of Closed and Open Source Software Development

Researchers in the field of software engineering economics have associated team factors, such as team size and team experience, with productivity and quality. Since distributed and open source development have gained significance in the past few years, further empirical investigation is needed. Our study contributes to the empirical body of knowledge by addressing this development. In particular, we investigate the association between team factors and failure processing efficiency for closed source software releases of a large commercial software vendor and for open source software projects registered with SourceForge.net. We find significant links between team experience and the failure processing efficiency. However, our data does not show any evidence for adverse effects of distributed development. Our results further suggest that service level agreements and process governance are good tools to guarantee satisfactory processing times.

Conformance quality and failure costs in the software Industry: An empirical analysis of open source software

The quality cost concept is well known in production economics. Recently, it has received a lot of attention in the field of software engineering. However, empirical studies of the association between failure costs and conformance quality have only been conducted for closed source software projects, but not for open source projects. This paper addresses this research gap. On the one hand, our analysis revalidates findings from production economics. On the other hand, it extends the limited empirical knowledge in the software quality cost research domain.

Analysis of software quality cost modeling’s industrial applicability with focus on defect estimation

The majority of software quality cost models is by design capable of describing costs retrospectively but relies on defect estimation in order to provide a cost forecast. We identify two major approaches to defect estimation and evaluate them in a large scale industrial software development project with special focus on applicability in quality cost models. Our studies show that neither static models based on code metrics nor dynamic software reliability growth models are suitable for an industrial application.

A Decision Support Scheme for Software Process Improvement Prioritization

Software managers pursuing process improvement initiatives are confronted with the problem of selecting potential improvements. In the field of software quality assurance, suitable decision support for prioritizing the optimization of activities according to their return on investment is not yet available. Our paper addresses this research gap. We develop a decision support scheme that facilitates the selection and prioritization of quality assurance activities. We demonstrate the scheme’s applicability in three industrial case studies. By relying on the well-known COQUALMO model’s characteristics and calibration data, our approach is industrially applicable with little data collection efforts.