This work in progress paper will investigate the effect of multiple deployment strategies for MATLAB® Autograder on student performance. Securing the resources required to adequately assess students in large class settings is a common problem faced by many universities and particularly first-year programs, where large lectures are common. These courses require the ability to quickly and efficiently assess and return student work, to optimize learning and satisfy an ever-increasing student desire for instant feedback. The need to achieve these ends in a cost-efficient manner has led to the development and adoption of auto-grading systems in many coding courses. While auto-graders serve as a useful tool in reducing time spent grading by instructors and improving response time for students, it has been noted this advancement can come at the expense of student-instructor interaction and the level of detail in feedback for students. Specifically, auto graders are primarily good at quickly assessing the correctness of answers while not dealing well with nuance or offering students detailed feedback on what specifically they can do differently. Careful development and deployment of assessment strategies is thus critical to maximize student outcomes while minimizing overall cost.
In this work-in-progress study MATLAB® Grader will be used to assess first-year student performance using multiple deployment and grading strategies while student outcomes and attitudes are tracked. Specifically, graded content will be delivered to students using 2 primary modalities. In the first, students will be delivered assignments that represent a relatively small portion of their grade (10% during the course of the semester), which will have a “unlimited attempt” grading strategy whereby students are free to repeat their assignments as many times as necessary in order to get the answer correct. In the second modality, students will be given an identical set of assignments weighted as 25% of their grade, with a limited number of attempted submissions to the auto-grader. Outcomes will be assessed for both student groups using common examinations in order to determine which strategy optimizes individual student performance.
Are you a researcher? Would you like to cite this paper?
Visit the ASEE document repository at
for more tools and easy citations.