Evidence-based software engineering

Abstract

Objective: Our objective is to describe how softwareengineering might benefit from an evidence-basedapproach and to identify the potential difficultiesassociated with the approach.Method: We compared the organisation and technicalinfrastructure supporting evidence-based medicine (EBM)with the situation in software engineering. We consideredthe impact that factors peculiar to software engineering(i.e. the skill factor and the lifecycle factor) would haveon our ability to practice evidence-based softwareengineering (EBSE).Results: EBSE promises a number of benefits byencouraging integration of research results with a view tosupporting the needs of many different stakeholdergroups. However, we do not currently have theinfrastructure needed for widespread adoption of EBSE.The skill factor means software engineering experimentsare vulnerable to subject and experimenter bias. Thelifecycle factor means it is difficult to determine howtechnologies will behave once deployed.Conclusions: Software engineering would benefit fromadopting what it can of the evidence approach providedthat it deals with the specific problems that arise from thenature of software engineering.