We recorded the percentage of students voting on each option on every clicker question used in 19 introductory statistics classes. To identify the questions that provide the most useful feedback, we looked for those for which the percentage of students voting correctly varied the most from class to class, producing the largest standard deviations. We present the nine questions with the highest standard deviations, discuss how students voted on them, and explain how we used them. We find these questions to be very diverse, but possessing especially well-designed distractors.