bulosehi

Billgates,
Regarding your first point - the notion of circumstantial evidence doesn’t really come into play here. Since this is a civil matter, the burden of proof the CFAI legally has to meet is not reasonable doubt, rather it is on a balance of probabilities. In order for the CFAI to sanction you and not be subject to legal recourse (i.e. a lawsuit), they must be reasonably satisfied that the similarities between the two answer sheets is more likely (ie: 50%) to be the result of cheating than not.
That said, this argument applies to both parties involved. There should be two investigations going on.
I can’t speak to the possible penalties. Good luck.

bulosehi

I wonder what level of significance their statistics are set for? There’s something like 100,000 candidates these days, or is it 200,000.
A 5% alpha level would imply that 2500 - 5000 people might be false positives (assuming a single tail). A 1% alpha level would imply 500-1000 people might be false positives.
It’s fine to use statistics for screening for possible cheating, but there’s got to be other types of evidence before one can summarily suspend people.
I understand that this is a private organization and so CFAI can more or less do what they want, applying whatever standards they feel, but surely it can’t be good to say… Please take our test, and by the way, we’re going to throw about 2.5% of you out each year and accuse you of statistically cheating.