Personal software process: An experiential report

Author(s): Soheil Khajenoori and Iraj Hirmanpour
Venue: Proceedings of the 8th SEI CSEE Conference on Software Engineering Education
Date: 1995


This study focuses on presenting results from a PSP class given to 24 Motorola engineers plus post-training data provided by six volunteer engineers six months after the course. The results show that between programs 1-3 and programs 8-10 the total number of defects is reduced by 27%, compile defects are reduced by 61%, and test defects are reduced by 63.2%. Additionally, based on the data presented by the six engineers, six months after the course the engineers make improvements of an additional reduction of 60% in both total defect density and test defect density. This means that after the additional time to adjust to the PSP process to their individual programming behavior, engineers reduced their total defect density by 86.6% and test defect density by 85.7% from when no PSP processes were used.

This study provides convincing evidence that the PSP process can greatly reduce the number of defects that exist in software. If one accepts the premise that the fewer defects found during development reflects the total number of defects in the product then these reductions in the number of defects represent an overall improvement of the software.

While this data shows that PSP can be used in industry and greatly reduce the number of defects in the software, it is important to recognize that the data was collected by the individual engineers and therefore may not accurately represent the actual effect of PSP. This study also has no control group so it is difficult to discern whether giving engineers training in other processes could create the similar results in defect reduction.