From: Dennis Lawrence <getafix@mindspring.com>
Date: 10 Mar 2000
Time: 22:01:54
I have a question on determining sample size for the following situation. 1. An organization is developing some computer software. They produce the usual artifacts – requirements specification, design specs, code. 2. There is a review organization that reviews all the documents, including code, for correctness. The intent is to find faults prior to release of the software. 3. There is an audit organization that will check up on the reviewers. There are many aspects to auditing, including the following. 4. One of the things the auditors will be looking for (in order to judge the quality of the reviewers) are pervasive errors by the reviewers. There are several methods that will contribute to a conclusion on this. One of these is to examine a sample of the developers documents, comparing the auditors results with the reviewers results.
The questions are: (1) How large a sample should be used? (2) Are there references that justify this sample size?