Empirical Research Methods in Software Engineering

Abstract

Software engineering is not only about technical solutions. It is to a large extent also concerned with organizational issues, project management and human behaviour. For a discipline like software engineering, empirical methods are crucial, since they allow for incorporating human behaviour into the research approach taken. Empirical methods are common practice in many other disciplines. This chapter provides a motivation for the use of empirical methods in software engineering research. The main motivation is that it is needed from an engineering perspective to allow for informed and well-grounded decision. The chapter continues with a brief introduction to four research methods: controlled experiments, case studies, surveys and post-mortem analyses. These methods are then put into an improvement context. The four methods are presented with the objective to introduce the reader to the methods to a level that it is possible to select the most suitable method at a specific instance. The methods have in common that they all are concerned with quantitative data. However, several of them are also suitable for qualitative data. Finally, it is concluded that the methods are not competing. On the contrary, the different research methods can preferably be used together to obtain more sources of information that hopefully lead to more informed engineering decisions in software engineering.