Just a short post to point to a newly added paper to the reference section. A few years ago I recalled seeing a paper that studied the difference to expect between various parts count methods and actual results.
Jeff and colleagues did this work some time ago, and in most cases the underlying parts count methods haven’t changed too much, so I suspect the results are still very relevant.
The bottom line – expect as much as -100% to +500% different between the prediction and the actual result.
To see a draft of the paper, visit the References section of the site, click Draft Comparison of Electronic Reliability Prediction Methodologies, or via the slideshare window below.
Enjoy. And, if you know of other studies of this nature. Please let me know and we’ll at least post a link to the work.