Students registered in numerical-based problem-solving courses are often given a number of assignments to complete independently in order to demonstrate and refine their problem-solving skills. Traditionally, these assignments are paper-based and all students receive the same problems to solve; thus, they often rely heavily on their peers or on solution manuals to complete their assignments. As a result, assignment grades are typically high, but do not correlate with test or exam performance.
 In this paper, we describe the use of Numbas, an open educational resource created by the University of Newcastle, England, as a customizable, online assignment system. Using Numbas, each student is provided with a unique set of problems, each with randomly generated values. While they are still allowed to work collaboratively with their peers, this randomization encourages students to develop their critical thinking skills to solve unique problems. To identify if the use of the online assignment system is correlated with enhanced performance, final exam grades earned by students who were exposed to either the paper-based or the online assignment system were compared.
 Furthermore, data from student feedback surveys were analyzed to identify student-perceived strengths and challenges associated with the online assignment system, and to determine possible opportunities for improvement. The study demonstrated an improvement in knowledge-based skills among students who were exposed to the online assignment system, compared to those who wrote paper assignments. However, no significant improvement in problem-solving skills was observed. Similar findings have been reported by other research works studied the same concept. Further, 88% of students surveyed reported that the online assignment system improved their learning experience.