And I've always wondered something about curves: Do they just look at how bad the entire test taking population did overall and then make the curve so that it gives a number the test takers think is appropriate for giving 5,4,3,2,1 to? And how does this account for better performance on tests themselves which is hidden by a more difficult test?
About six questions on the test are "trial" questions. The AP Exam doesn't really count them for the score. In addition, yes the exam is curved by semi-mysterious ways, but the grading of the free-response questions is an exact science. They have really well-thought out rubrics.
High School Computer Science - Community to helping students pass the AP Exam in Java.