Abstract
In this dissertation we propose the Mastery Assessment Pedagogy for Learning Computer Science, or M.A.P. Learning, an approach to assessment inspired by the original Mastery Learning approach which takes into account more recent developments in the learning sciences as well as issues specifically related to the field of CS. We provide an overview of the design of the approach and data from three studies in which the approach was used. M.A.P consists of four core principles intended to fulfill the spirit of Mastery Learning while taking advantage of research on the practice which has been published in the past few decades. The principles are as follows: Individual Assessment on a Discrete Set of Skills, High-Frequency/Low-Stakes Assessment, Secure and Robust Assessment, and Pass/Fail Grading. In practice, these principles are implemented as weekly mastery assessments on a set of core skills which are graded pass/fail. A new assessment is provided each week for every skill which has been introduced in the course so far, and students are given the choice of which and how many skill assessments they would like to attempt that week. Once a student has demonstrated mastery over a skill, they are no longer required to attempt assessments for it, and their final grade is calculated based on the number of skills they mastered during the semester. Variations of M.A.P. were used in a Discrete Math course, a Software Engineering course, and two semesters of CS1. This dissertation demonstrates that M.A.P. successfully supports the goals and spirit of Mastery Learning, while adjusting features which made the original approach difficult to implement.