The Personal Software Process: Experiences from Denmark

Author(s): Pekka Abrahamsson and Karlheinz Kautz
Venue: Euromicro Conference
Date: 2002

Sample Size: 22
Class/Experience Level: Undergraduate Student
Participant Selection: class participation
Data Collection Method: Project Artifact(s)


Type of Experiment: Experience Report
Sample Size: 22
Class/Experience Level: 4th and 5th year students
Participant Selection: Participants in a PSP course
Data Collection Method: PSP data logs

This is a report that describes the experiences of a class of students taking a PSP course. The authors start with a brief overview of the PSP method and then discuss the results of their analyses on the PSP data that was collected by the students during the class. The students’ ability to estimate program size didn’t improve during the course; however it did stabilize such that the estimation error range was predictable. Their data also showed that once students reached the level of PSP where design and code reviews were introduced, the number of defects in the compile and test phases decreased.

Data collection was performed manually, but an extra step of data validation was performed to try and keep the quality of data high. This data validation was performed because a known problem with the PSP process is the significant impact that bad data can have on the results, and the relatively high occurrence of data collection problems.

An interesting feature of this report is the section that includes feedback comments from the students. Feedback from the students indicated that they were surprised at how often they were interrupted while working on a task, and that they hadn’t been aware of this until they began recording PSP metrics. The student’s came away with the impression that the process had its merits, but was very time consuming and would benefit from automation.