Web-based automatic feedback on assignments in statistics: How can it help students learn statistics and universities reduce costs?

A non-experimental study in 2005 suggested that immediate, automatic feedback on assignments helped to increase study motivation as well as pass rate among engineering students attending an introductory course in statistics at Oslo University College. In the follow-up study reported here we used an experimental design assigning the participants randomly to one of two experimental conditions: The “web-supported” students received immediate, automatic feedback after having entered their responses to the assignments electronically. The “paper-supported” students received written feedback on their paper-based submissions several days later. The findings contradicted the results of the non-experimental study: No significant differences between the groups were found with regard to final examination grades, study effort (with a certain reservation), and preferences with regard to the method for submitting answers. Running tutoring costs, however, were much lower for the web-supported than for the paper-supported students. Closer analysis of the results suggests that the actual benefits of web support relative to paper support in the reported experiment may include improved learning outcomes in addition to the reduction of tutoring costs. Also, observations of student work habits indicate that the learning outcomes stimulated by the learning-support system can be improved by inexpensive modifications compelling the students to work more regularly and perform activities that may strengthen learning. For example, the students may be obliged to submit electronically answers to more problems than demanded in the reported experiment. Further, in order to increase the pressure on the students to think and understand before they submit the answers, we may require the students to supplement their web-submissions of answers by handing in paper-based documentation of the reasoning and the calculations underlying these answers. The efficacy of such modifications of the learning-support system needs to be tested experimentally. Overall, our studies illustrate the value of pursuing a persistent process of research-based, step-wise, and long-term improvement of the tools and the methods used to promote student learning.