Calibrated Peer Reviews in Requirements Engineering Instruction: Application and Experiences

Abstract

Instructing Requirements Engineering (RE) is a challenging task due to the absence of single absolute and correct solutions computer science students so often strive for. Instead, there is often a variety of compromise solutions for each RE problem. Therefore, it is essential that aspiring Software Engineers are exposed to as many solution alternatives as possible to experience the implications of RE decisions. To facilitate this, we propose a learning-by-multiple-examples process, in which we make use of a calibrated peer review grading model for assignments. Paired with a think-pair-share model of semester-long, industry-realistic, project-based low-stakes milestones, we were able to generate a rich collaborative learning atmosphere. In this paper, we report the course design and experiences from the application of calibrated peer reviews in an undergraduate RE course. Qualitative and quantitative application results show that calibrated peer reviews significantly improve students’ learning outcomes.