Investigating the applicability of the evidence-based paradigm to software engineering

Abstract

Context: The success of the evidence-based paradigm in other domains, especially medicine, has raised the question of how this might be employed in software engineering.Objectives: To report the research we are doing to evaluate problems associated with adopting the evidence-based paradigm in software engineering and identifying strategies to address these problems.Method: Currently the experimental paradigms used in a selected set of domains are being examined along with the experimental protocols that they employ. Our aim is to identify those domains that have generally similar characteristics to software engineering and to study the strategies that they employ to overcome the lack of rigorous empirical protocols. We are also undertaking a series of systematic literature reviews to identify the factors that may limit their applicability in the software engineering domain.Conclusions: We have identified two domains that experience problems with experimental protocols that are similar to those occurring for software engineering, and will investigate these further to assess whether the approaches used to aggregate evidence in these domains can be adapted for use in software engineering. Our experiences from performing systematic literature reviews are positive, but reveal infrastructure problems caused by poor indexing of the literature.