They are probably reporting the expected rate based on the applicable population, not for those who study. I know 3 people who took the '1994 test; one who had good experience and studied, one a loud mouth who studied, and the third was a loud-mouth know-it-all who didn't need to study. Of them, only the last didn't pass the first time, so that's a 66% pass rate in that tiny sample.
Based on the work of those who haven't studied, I'd say 15% is high, but maybe TecEase finds those who have self-selected.
What is more key is if a lot of the test is 'parrot.' This is the type where phrases are slightly altered from the exact wording of the standard and it is up to the test taker to identify which most accurately parrots the original. 'Parrot' questions require memorization but not much understanding. Taking off points, for example, by not recognizing that a datum cannot be modified; that modifiers only apply to datum features.
I contrast those with 'ferret' questions, where one needs to understand how the rules work to produce results.
'Parrot' questions are popular because they are easy to make into multiple choice questions and true-false questions.
Some are likely to be hard. For example, on
, question 3 is a question that is not answerable because the relation between the involved datums -ooops- datum features, isn't specified. Answering it requires making an assumption about the relationship between datum feature B and datum feature A.