<p>@Quantmech</p>
<p>Your evil plan raises an interesting question: does the distribution of answers change as the test questions get harder? Maybe if Fignewton still has his data set, he can break it down…</p>
<p>I’m guessing the answer is no. But if the answer is yes, it doesn’t necessarily indicate evilness. They just may not want students getting the right answer for the wrong reason. You can accidentally write a question that has weaker students getting it right more often than stronger ones so that the testers scoring 500 and 750 get it right but the kids in the 600s miss it.</p>