A Holistic Overview of Software Engineering Research Strategies

Abstract

Empirical research studies are the principal mechanism through which the software engineering research community studies and learns from software engineering practice. The focus on empirical studies has increased significantly in the past decade, more or less coinciding with the emergence of evidence-based software engineering, an idea that was proposed in 2004. As a consequence, the software engineering community is familiar with a range of empirical methods. However, while several overviews exist of popular empirical research methods, such as case studies and experiments, we lack a 'holistic' view of a more complete spectrum of research methods. Furthermore, while researchers will readily accept that all methods have inherent limitations, methods such as case study are still frequently critiqued for the lack of control that a researcher can exert in such a study, their use of qualitative data, and the limited generalizability that can be achieved. Controlled experiments are seen by many as yielding stronger evidence than case studies, but these can also be criticized due to the limited realism of the context in which they are conducted. We identify a holistic set of research methods and indicate their strengths and weaknesses in relation to various research elements.