Innovating assessment at the University of Nevada, Las Vegas!

Streamlined Integration and Practical Learning

UNLV’s Computer Science courses are seamlessly integrated with the WebCampus/Canvas system, allowing a smooth transition to CodeGrade for both instructors and students. Courses like CS I, which caters to 50-60 students per section and involves 8-12 projects per semester, heavily rely on practical, hands-on learning. CodeGrade facilitates this by enforcing thoughtful code corrections through a structured resubmission process and imposing penalties for late submissions, ensuring students stay on track.

Scalable Grading and Personalized Feedback

In courses with large enrollment numbers, such as CS I, which can have up to 1,000 students per year, manual grading is impractical - over 20,000 submissions! Alex and Ed automate this process, allowing them to focus on providing detailed, line-by-line feedback rather than getting bogged down by the sheer volume of submissions. This not only improves the consistency of grading but also enhances the learning experience by enabling personalized feedback on every submission.

CS I is a required course for many non-CS majors, so giving enough practice opportunities is paramount. Learning to code can be intimidating - Ed and Alex stress that by giving several small assignments related to that week’s curriculum, new students can improve their confidence. They also require that teaching assistants leave an encouraging feedback comment on every submission.

Alongside large-scale introductory courses, Ed Jorgensen also teaches advanced coding courses with CodeGrade. Systems Programming covers Intel x86 assembly language, which runs natively, and RISC MIPS assembly language, executed via a simulator. Students use CodeGrade’s unit testing feedback and resubmission features to master these unfamiliar topics.

Another use case is for their Programming Languages course. Students work on rotating projects in various languages, including Perl, Bash, Python, Ruby, Fortran, Java, LISP, and Gnu Prolog. The course requires setting up compilers or interpreters in CodeGrade. More senior students make better use of testing and resubmission opportunities.

Research-Driven Results

Alex conducted research comparing automated grading with human-based grading, using his weekly lab sessions as the base. Here's how it was set up:

• Students were given 2 weekly programming lab assignments
• Half of the students were traditionally graded by humans, the other half were subjected to automatic grading using CodeGrade
• After each week grading types were swapped between the two groups
• 10 weeks of labs were given (20 labs in total)

The results showed that CodeGrade significantly reduced grading variability, which often occurs due to fatigue in manual grading. The tool's consistency was particularly beneficial in large classes, where maintaining uniformity across thousands of submissions is challenging!

Supporting a Diverse Student Body

UNLV’s CS department has seen significant growth, particularly among students from underrepresented groups and those with no prior background in computer science. CodeGrade plays a crucial role in supporting these students by providing instant feedback, allowing them to learn from their mistakes and improve continuously. This has contributed to increased retention rates, as students are more engaged and can make corrections promptly.