LAUSR

Software engineers must solve practical problems under deadline pressure. They rely on the best-codified knowledge available, turning to weaker results and their expert judgment when sound science is unavailable. Meanwhile, software engineering researchers seek fully validated results, resulting in a lag to practical guidance. To bridge this gap, research results should be systematically distilled into actionable guidance in a way that respects differences in strength and scope among the results. Starting with the practitioners’ need for actionable guidance, this article reviews the evolution of software engineering research expectations, identifies types of results and their strengths, and draws on evidence-based medicine for a concrete example of deriving pragmatic guidance from mixed-strength research results. It advances a levels-of-evidence framework to allow researchers to clearly identify the strengths of their claims and the supporting evidence for their results and to work with practitioners to synthesize actionable recommendations from diverse types of evidence. This article is part of a special issue on software engineering’s 50th anniversary.