Students as Study Subjects in Software Engineering experimentation

Abstract

Experimentation using different types of subjects is an important issue in empirical software engineering. In particular, the use of students as subjects is many times questioned. This paper addresses this issue by dividing subjects into four types, where one type can be viewed as a worst case. It is discussed how students could be viewed as suitable subjects in the worst case. In particular, if it is possible to reject the null hypothesis then it is highly likely that the outcome is also valid for other types of subjects. An example experiment is presented that uses students as subjects and fulfill the criteria for being the worst case. It is concluded that students are good subjects under certain circumstances and that these types of experiments provide an opportunity for using students as subjects. 1.