Introducing the personal software process

Author(s): Watts S. Humphrey
Venue: Annals of Software Engineering
Date: 1995


The results presented in this study were conducted before the SEI published the Personal Software Process and were based on data gathered from teaching the process to graduate students at five different universities. The data shows that there can be significant variation in the percent reduction of total defects when applying PSP. Programmers who are experience software engineers have a reduction in total defects between 45.8% and 55.1% while students with little experience in software engineering have a reduction in defects between 64.0% and 80.1%. These results suggest that the PSP process in more effective at reducing defects when a developer is less experienced. The study also looks at productivity in terms of lines of code per hour and finds that productivity improves between 22.9% and 136.0%. Unlike defect reduction, there is no correlation between productivity and the software engineering experience of the developer.

This study provides the initial evidence that PSP can significantly reduce the number of defects in programs without decreasing productivity. Like many studies conducted on PSP, this study relies on the data collected by the individual participants which may be prone to errors because of difficulties in controlling the accuracy of data collection. Additionally, some argue that studies that use the ten programs in the PSP training curriculum to evaluate the effect of the process may inaccurately reflect reality because of the nature of the programs and reuse of code from one program to the next.