DK,

Unfortunately, fraud is a concern. We would require some sort on monitor to verify the test environment and test taker.

I agree with your approach - asking for significant numbers from an analysis to confirm that the model was built and analyzed "correctly".

Maybe we could establish two tests - the official certification exam and a second (free?) self-assessment where users can run through a series of questions on their own. But there I go; we don't even have a test built and already I'm talking about having two!

Thank you for your thoughts.