Introductory programming courses rely heavily on using programming assignments to help teach students the basics of developing software.
There exists a need to assess these assignments automatically to reduce the strain caused by manual inspection on limited course staff resources.
Learning to program includes software testing.
Taditionally automatic assessment systems have relied on code coverage metrics to assess the test suites generated by the students.
This work demonstrates weaknesses of this approach and proposes mutation testing as an alternative.
Mutation analysis tool Javalanche is evaluated and used on actual coursework in the university's programming courses.
The results are analyzed quantitatively and qualitatively to demonstrate the strengths and weaknesses of the new approach.
