Evaluating Software Design Processes by Analyzing Change Data Over Time.

This paper presents analyses of early design and code change data from the Software Cost Reduction (SCR) project, a well- reported effort conducted at the Naval Research Laboratory from 1978 to 1988. The analyses are mostly time-based studies of the change data and relationships between the data and SCR personnel activity data. This analytical approach seems to allow useful insights into software design processes even when data are limited to a single software project. It also enables project personnel to notice favorable or unfavorable patterns with respect to project goals during the course of the project. Some analyses of the change data show patterns consistent with a major goal of the SCR project—the design and development of easy-to-change software. Specifically, most changes took a day or less to uncover and resolve; the majority of changes updated at most one mod- ule. Moreover, these percentages remained fairly stable. Also, no positive relationship appeared between error-correction effort and the number of days that an error remained in the SCR design documentation. Other analyses suggest that consistency may have been temporary. For example, the analyses suggest a stepwise growth in average change effort, and an increasing percentage of changes resulted in module interface updates. Certain specific ratios between SCR change data and personnel activity data show promise as possible indicators of design incompleteness. The ratios are based on data of the kinds that are typically collected on software projects. [ABSTRACT FROM AUTHOR]