From Multiple-Choice Through Free-Text Submissions And Qualitative Peer-Reviews To Grading Students And Reviewers

Methods for peer-evaluation of free-text submissions were tested in three courses, one a theoretical course on linear models (for graduate students in statistics), one on technical aspects of cryptocurrencies (for computer science and mathematics students) and one course offering remedial mathematics for graduate students working towards a degree in applied statistics. Quantitative and qualitative assessments were obtained, both for the student submitter and the peer-reviewer.Qualitative results are given at several levels regarding the different reporting tendencies of students (level of elaboration of submissions) and reviewer (quality of feedback). Quantitative results are provided on the consistency of reviews, including comparisons, both internally and with other methods of assessment, from large end-of-course projects to final exams.Approaches for counteracting grade inflation are introduced; their potential efficacy discussed and their quantitative evaluations are outlined.The studies were implemented using a revision of the multipurpose tutor-web tool which doubles as a learning management system and a research toolbox. The revised tool includes several options for student submissions of free text, ways to implement peer-review and grading schemes, both peer-grades for submissions and grades for the review process itself.In addition to grades the tutor-web allocates cryptocurrency rewards for performance using the Smileycoin. An implementation of this in the context of free-text submissions and peer-review is discussed.