Game-theoretic models identify useful principles for peer collaboration in online learning platforms Conference Paper


Author(s): Pandey, Vineet; Chatterjee, Krishnendu
Title: Game-theoretic models identify useful principles for peer collaboration in online learning platforms
Affiliation IST Austria
Abstract: To facilitate collaboration in massive online classrooms, instructors must make many decisions. For instance, the following parameters need to be decided when designing a peer-feedback system where students review each others' essays: the number of students each student must provide feedback to, an algorithm to map feedback providers to receivers, constraints that ensure students do not become free-riders (receiving feedback but not providing it), the best times to receive feedback to improve learning etc. While instructors can answer these questions by running experiments or invoking past experience, game-theoretic models with data from online learning platforms can identify better initial designs for further improvements. As an example, we explore the design space of a peer feedback system by modeling it using game theory. Our simulations show that incentivizing students to provide feedback requires the value obtained from receiving a feedback to exceed the cost of providing it by a large factor (greater than 7). Furthermore, hiding feedback from low-effort students incentivizes them to provide more feedback.
Keywords: Game theory; MOOCs; Online collaboration; Peer feedback
Conference Title: CSCW: Computer Supported Cooperative Work and Social Computing
Volume: 26-February-2016
Conference Dates: February 26 - March 2, 2016
Conference Location: San Francisco, CA, USA
ISBN: 978-145033950-6
Publisher: ACM  
Date Published: 2016-02-27
Start Page: 365
End Page: 368
Sponsor: ERC Start Grant Graph Games 279307 supported this research.
DOI: 10.1145/2818052.2869122
Open access: no
IST Austria Authors
Related IST Austria Work