Online programming exercises for summative assessment in university courses

Although interactive e-learning environments are increasingly used in university courses, traditional types of examination still dominate the way how students are assessed for grades. In this paper, we examined how student data from online interactive learning environment for programming exercises can be used for summative assessment at the end of the course. Using data from three different university courses, we calibrated the two parameter logistic regression model, ranked students according to their ability of solving problems, and matched them to final grades. Results indicate we can predict grades within 0.57 to 1.02 level of accuracy, suggesting that careful use of interactive e-learning environments in university courses can substitute existing assessment methods, opening further possibilities for innovation in instructional process.