methods that are based on these metrics, such as cost estimation, seem not to have been appreciated. Subset selection methods (e.g., Mallows, 1973) provide one way to assess variable redundancy and the effect on fitted models, but other approaches that use judgment composites, or composites based on other bodies of data (Tukey, 1991), will often be more effective than discarding metrics.

Metrics typically involve processes or products, are subjective or objective, and involve different types of measurement scales, for example, nominal, ordinal, interval, or ratio. An objective metric is a measurement taken on a product or process, usually on an interval or ratio scale. Some examples include the number of lines of code, development time, number of software faults, or number of changes. A subjective metric may involve a classification or qualification based on experience. Examples include the quality of use of a method or the experience of the programmers in the application or process.

One standard for software measurement is the Basili and Weiss (1984) Goal/Question/ Metric paradigm, which has five parameters:

  1. An object of the study—a process, product, or any other experience model;

  2. A focus—what information is of interest;

  3. A point of view—the perspective of the person needing the information;

  4. A purpose—how the information will be used; and

  5. A determination of what measurements will provide the information that is needed.

The results are studied relative to a particular environment.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement