Measuring (and enhancing?) student confidence with confidence scores

The important skill of building confidence in one's analysis through sanity and cross-checking is often poorly acquired by engineering students. An introductory circuit analysis class presents an ideal opportunity in which to emphasize and measure this skill, since problems can typically be worked with a number of different methods or worked "backwards" to provide cross-checks. This paper reports on a two-semester experiment in which students were required to provide a confidence rating for all exam and quiz answers. The structure allowed for expressing positive confidence (confidence that an answer is correct), negative confidence (confidence that an answer is incorrect) and neutral confidence. A confidence score that measured how well the students evaluated the correctness or incorrectness of their answers was combined with a traditional problem score to form the exam score. We present numerical results of this experiment, which yield potentially valuable conclusions regarding students' perceptions of the correctness of their answers.