A replication empirical study of training, experience and software review performance

Recent research suggests that task training experience has not significant effect on software review performance. The study directly contradicts the software review literature. As a result, the aim of this paper is to conduct a replication experiment to validate the important relationship between training, experience and software performance. A laboratory experiment was conducted in winter 2003. One hundred and five volunteer university students were employed. Subjects were required to detect defects from a requirement document. The main findings include (1) trained subjects did not find more true defects than non-trained subjects, but they found less false positives than the non-trained subjects; (2) trained subjects who have role experience (i.e. software review experience) did not perform better than non-trained subjects in the software review task; (3) trained subjects who have working experience in software industry performed a little better than non-trained subjects.