The purpose of this paper is to present results of a recent study in which students voted on multiple choice questions in mathematics courses of varying levels. Students used clickers to select the best answer among the choices given; in addition, they were also asked whether they were confident in their answer. In this paper we analyze data collected from 192 multiple-choice questions used in five courses (Precalculus, Calculus 1, Differential equations, Linear Algebra, and Real Analysis) by two instructors at two different institutions. We evaluate student votes using four categories: correct and very confident, correct but not very confident, incorrect but not very confident, and incorrect and very confident. We examine and discuss characteristics of some questions from each category. We further assert that access to confidence data of this type will be an effective planning tool for future classroom voting. "Supplementary materials are available for this article. Go to the publisher's online edition of PRIMUS to view the supplementary file."