A secondary analyses of Bradac et al. s prototype process-monitoring experiment

Abstract

We report on the secondary analyses of some conjectures and empirical evidence presented in Bradac et al. s prototype process-monitoring experiment, published previously in IEEE Transactions on Software Engineering. We identify 13 conjectures in the original paper, and re-analyse six of these conjectures using the original evidence. Rather than rejecting any of the original conjectures, we identify assumptions underlying those conjectures, identify alternative interpretations of the conjectures, and also propose a number of new conjectures. Bradac et al. s study focused on reducing the project schedule interval. Some of our re-analysis has--considered improving software quality. We note that our analyses were only possible because of the quality and quantity of evidence presented in the original paper. Reflecting on our analyses leads us to speculate about the value of descriptive papers --that seek to present empirical material (together with an explicit statement of goals, assumptions and constraints) separate from the analyses that proceeds from that material. Such descriptive papers could improve the public scrutiny of software engineering research and may respond, in part, to some researchers criticisms concerning the small amount of software engineering research that is actually--evaluated. We also consider opportunities for further research, in particular opportunities for relating individual actions to project outcomes.