Personalized feedback emails: A case study on online introductory computer science courses

Abstract

The absence of face-to-face interaction between instructors and students in online courses has been the focus of discussion in many research papers. To compensate for this defect, the concept of Personalized Feedback Email (PFE) was introduced in two undergraduate online courses at the University of Georgia. A distinct component of PFE is a grade forecast for each individual student projected visually in graphs. The quantitative and qualitative data collected from students made it possible to claim that PFE contributes to students' engagement in online courses and encourages the majority of them to do better in class. Given that the rate of contribution of each student in course activities is correlated with student's performance, we were able to show that students who find PFE motivating make higher contributions in class activities. PFE is especially capable of targeting students who stand in the middle of the grade-range and improves their contribution and performance. In this respect, PFE also has a considerable short-term effect. The extensive applications of this effect should be limited by the optimization of the number of PFEs. All this machinery is expected to enable the complex of decision-makers associated with students to adopt the most effective learning strategies. This study shows a drastic and positive change in the performance of students who alter their learning strategy after being exposed to their forecasted grades, which enhances the potential of supervised improvement. The accuracy of forecasting model will be crucial when forecast grades are expected early in the semester to identify at-risk students. Applying machine learning methods, particularly the Greedy Linear Regression, satisfies this expectation and increases the correlation coefficient of the forecasts to 0.98.