by esledge Tue Oct 27, 2009 2:50 pm
Our CAT algorithm, like the real GMAT is probablistic, not deterministic. That means that when you get a question right, the algorithm randomly selects your next question from the pool of new questions, with a higher selection probability for (1) harder questions (compared to the one you just got right), (2) question types you haven't seen enough of, (3) question topics you haven't seen enough of, etc. There is a lower selection probability for the easier questions or those with topics you have already been tested on.
In short, you don't always get a really hard question, even when doing really well.
It is also possible that since you were on test #5, the algorithm was running short of 700-800 questions to give you, especially if you have scored at a high level in all 5 exams. We generally tell people they can take 6 of our exams before the algorithm needs to recycle questions, but for people at the tail ends of the bell curve, that number of exams may be smaller.
Don't worry about the non-representative score--just look at it as extra practice on those almost-hard-enough questions. You WILL see those on your official exam, and you must nail them to get a top score!
Emily Sledge
Instructor
ManhattanGMAT