The collection and use of evidence in software engineering practice and research are essential elements in the development of the discipline. This paper discusses the need for evidence-based software engineering, the nature of evidence in its various forms and some of the research methodologies used in other disciplines for the collection of evidence, which are also relevant to software engineering. Two frameworks or models are proposed which illustrate the relationships between the methodologies discussed. In particular, the paper highlights the importance and roles of both positivist and interpretivist methods of investigation.