On tackling quality threats for the assessment of measurement programs: A case study on the distribution of metric usage and knowledge

Abstract

Context: Measurement programs are subject to changing requirements, causing the need for a robustness to compensate these changes. Existing methods enable the assessment of measurement programs and their robustness with the help of company representatives. Goal: However, it is not clear how much the spread of knowledge among roles and differences between company internal organizations influence the result and quality of the assessment. Thus, it is not known today who and how many representatives should be interviewed to get reliable results. Method: To address these points we performed a case study on the assessment method MeSRAM, spanning 18 interviews with 14 representatives of 4 different organizations within a large international communication company. Results: Managers are able to answer more questions and their answers are more often correct than those of engineers. Furthermore, most errors occur for questions about the use of concrete metrics, while for aspects about the metrics' infrastructure no conflicting answers occurred. We also identified 6 reasons for gaining incorrect answers. Conclusion: The assessment of measurement programs requires additional care to ensure the quality of the results. • A case study on the assessment of software measurement programs. • Investigation of quality threats to that arise due to knowledge distribution. • A first list of reasons for assessment mistakes and lessons learned.