Selection of Projects in Software Process Assessment: New Perspectives

Abstract

Since the starting point of successful software process improvement begins with a formal software process assessment, it is essential that this assessment is effective and reliable. However, recently, critics and proponents have engaged in a debate on the reliability of such assessments. Until more scientific assessment methods are ready, how can the current kind of assessments be made more reliable? This paper argues that it is possible to reduce variability of software process assessments and improve their reliability. An assessment is an intensive human process of decision taking, subject to different interpretations by the evaluators. To overcome this, automation and quantitative data that could help in decision taking are needed. This paper discusses the need for widespread participation by everybody from the organisational unit (the part of the organisation being assessed), its implications for project selection and proposes the use of an automatic method for a better project selection. Better selection of projects for the visit, and better information about the organisation, could imply a visit in less time and with fewer costs.