Usage 2: The Defect density is usually expressed in Major-Defects per Logical-Page.
EXAMPLE OF ESTIMATING REMAINING DEFECTS For example,
if the Defect-Detection-Process had a historic and stable Effectiveness of 80%, and
you found 80 Major-Defects, then
you might estimate that there are a total of 100 Major-Defects,
of which you found 80, and 20 Major-Defects you did not find.
If you correct the 80 you found then the
20 not found remain (plus or minus some uncertainty factor of about 30% of the total remaining).
If you historically fail to correct 25% of what you try to correct, then
to the 20 you did not find or fix
you need to Add 20 (25% of 80 correction attempts) you failed to amend correctly.
This gives a total estimate of 40 Major-Defects remaining (20 never found and 20 not fixed correctly),
After you find 80 and try to correct them.
This method works well regularly, and can be confirmed by an SQC Process that manages to find a predicted number of Defects (for example 80% of the 40 remaining in the example above, i.e. about 24 Majors), at various clients. It gives the right order of magnitude, even when there is little history, and Conditions are somewhat unstable. If this seems too complicated, then use the following rule of thumb: "For every Defect you find and fix there remain that many more in the System." If that doesn’t Work for you, try Gerald M. Weinberg’s Principle, (Psychology of Computer Programming): "There is always one more Bug remaining (even After you think you have removed the last Bug)."
This Concept entered by Kay Dudman.