Yue Gong, Joseph E. Beck, Neil T. Heffernan, Elijah Forbes-Summers
One of the common expectations of ITS designers is that students efficiently learn from every practice opportunity. However, when students are using an Intelligent Tutoring System, they can exhibit a variety of behaviors, such as “gaming,” which can strongly reduce learning. In this paper, we present a new approach to infer the impact of gaming on learning at the fine-grained level. We integrated a knowledge tracing model of the student’s knowledge with the student’s gaming state (as identified by our gaming detector). We found that when gaming, students learn almost nothing (on the order of one-twelfth to one-fiftieth as efficiently). A student’s gaming amount is associated with aggregate effects on his knowledge and learning, leading to less learning even in the practice opportunities where no gaming occurs. In addition, we found that students tend to game in those skills on which they have relatively low knowledge. Furthermore, we found that knowing the identity of the studentis more important than knowing the skill for predicting whether gaming will occur.
The final publication is available at Springer via https://doi.org/10.1007/978-3-642-13388-6_24