Addressing Bias to Improve Reliability in Peer Review of Programming Coursework

Bradley, Steven

Abstract

Peer review has many potential pedagogical benefits, particularly in the area of programming, where it is a part of everyday professional practice. Although sometimes used for formative assessment, it is less commonly used for summative assessment, partly because of a perceived difficulty with reliability. We explore the use of a hierarchical Bayesian model to account for varying bias and precision amongst student assessors. We show that the model is sound and produces benefits in assessment reliability in real assessments. Such analyses have been used in essay subjects before but not, to our knowledge, within programming.

Citation

Bradley, S. (2019, November). Addressing Bias to Improve Reliability in Peer Review of Programming Coursework. Presented at Koli Calling 2019, Finland

• Presentation Conference Type: Conference Paper (published)
• Conference Name: Koli Calling 2019
• Acceptance Date: Sep 11, 2019
• Online Publication Date: Nov 21, 2019
• Publication Date: Nov 21, 2019
• Deposit Date: Oct 30, 2019
• Publicly Available Date: Jan 28, 2020
• Pages: 1-19
• Book Title: Koli Calling '19 : proceedings of the 19th Koli Calling International Conference on Computing Education Research.
• DOI: https://doi.org/10.1145/3364510.3364523
• Public URL: https://durham-repository.worktribe.com/output/1141608