Students are often unaware of potential gaps in their knowledge while preparing for course exams until it is too late. As a result, we developed a web-based assessment tool called the Confidence-based Learning Accuracy Support System (or "CLASS"). CLASS collects and operationalizes student confidence to generate metrics that highlight content areas where students excel, where they falter and areas where their level of confidence does not match their demonstrated mastery (i.e., their perceptions of their abilities are not accurate). As a result of implementing CLASS within a large-enrollment (i.e., ~100 per semester) introductory physical geology course, student outcomes (e.g., performance, perception accuracy) improved. These first iterations of use, however, relied upon quizzes that drew questions at random (10 per attempt) from question banks that were built from a sequence of learning objectives that were determined and arranged by the instructor. While these results were promising, we began to wonder, what would happen if we gave students control of their formative assessment?

After a semester of development, in the Fall 2019 semester we provided students the ability to customize their use of CLASS via a new feature titled Dynamic Quizzing. Dynamic Quizzing allows students to generate personalized quizzes based on course learning objectives of their choosing. To inform selections, we provide students their demonstrated levels of performance, confidence, and how these values compare to one another (a metric called bias). We will discuss how students used the new feature including which topics they chose to study, when they chose to study, and whether or not they selected the topics for which they truly needed the most practice. Finally, we will discuss how these behaviors affected exam outcomes for different populations of students.