My students echoed Paul's students. They thought the multiple choice was much easier than expected, and that the free-response questions were also very doable. We came up with about three different block designs for #2 in class, overall not a very difficult question. #3 gave students who hadn't reviewed probability as much fits, and in my opinion was the 'most' difficult of the free response (for the students). My answers were a) .06894 and b) .71076. #4 and #5 were pretty straight forward two sample problems, very much like problems my students had had on tests the whole course. On the Investigative task we had some interesting discussions about best models. My students seemed to think (after placing the data in their calculators) that the exponential model (the ln model in the problem) 'fit' the data best, but most stated that a regression model on only cars from the model year 1988 would be a better tool for studying the price (possibly looking at milleage as the independent variable). Again lots of different answers, with good explanations by the kids.
Overall, I too think it was a fair test, and am curiously and anxiously awaiting the results in July.