Stott, Daniel R. (1989, ASQC) Data Systems Assurance, IBM Corp., Kingston, NY
The purpose of this paper is to show how Shewhart charts can be used in software development processes, bringing the power of statistical thinking to people with limited knowledge of, or interest in, statistical methods. In this role, Shewhart charts will be critically important in the continuous improvement of software processes as we enter the nineties.
The software development "process" is the system of people, tools, and procedures that produce a software product. Historically, one measure of the performance of this process has been the number of defects found in a product by customers after shipment. In order to ensure product success in this area, software developers need to detect and correct deviations from expected process performance (control), and to prevent deviations from occurring in the future (improvement). Shewhart charts can satisfy this need. This paper identifies some metrics that measure the effectiveness and efficiency of the software development process, demonstrates how Shewhart charts have been used to analyze them, and highlights some of the conclusions that can be reached.
First, we show how a process capability study can be used to remove assignable causes of variation to produce predictable programming process performance. We then show how the resulting predictions of process performance can be used for process control. Finally, we show how Shewhart charts can be used during process postmortem studies to focus attention on critical areas.
Software quality,Shewhart control chart,Control charts