The Impact of Incorporating Student Confidence Items into an Intelligent Tutor: A Randomized Controlled Trial.
Charles Lang, Neil Thomas Heffernan, Korinn Ostrow, Yutao Wang · Educational Data Mining · 2015
For at least the last century researchers have advocated the use of student confidence as a form of educational assessment and the growth of online and mobile educational software has made the implementation of this measurement far easier. The following short paper discusses our first study of the dynamics of student confidence in an online math tutor. We used a randomized controlled trial to test whether asking students about their confidence while using an Intelligent Tutor altered their performance. We observe that (1) Asking students about their confidence has no statistically significant impact on any of several performance measures (2) Student confidence is more easily reduced by negative feedback (being incorrect) than increased by positive feedback (being correct) and (3) confidence accuracy may be a useful predictor of student behavior. This paper demonstrates how psychological ideas can be imported into Educational Data Mining and our findings point to the possibility of using student confidence to better predict performance and differentiate between students based on the way they approach items.