Peer assessment to improve reproducibility of computational project work

Computational research requires increased reproducibility for open science practices yet is not widely taught in the geosciences. Teaching reproducibility and establishing it as learning qualification and objective is an important step towards improving scientific practice and will also improve student learning by facilitating peer-review and reuse of earlier work in later courses. It will also improve the formal (summative and formative) assessment of student project work within a course, and the quality of reusable open educational resources. There is evidence in the literature that teaching reproducibility should combine practice (tutorials) trying to reproduce someone else’s work, and iterations of teacher and peer feedback on the
reproducibility of one’s own work. In this contribution, reproducibility was introduced as a new topic in a 15 EC MSc course, which follows a challenge-based learning approach to tackle the wicked problem of different stakeholders facing human-induced earthquakes due to gas extraction. Students work in groups for different stakeholders. Self-regulated feedback is encouraged to include other stakeholders’ views. After a tutorial at the end of the first half of the course, student groups submitted a reproducibility plan for their project work, which was then peer-reviewed by the other groups, so that any feedback could still be incorporated. The quality and depth of the peer feedback itself provided information on how well the topic has been understood. The outcomes show that the approach delivered encouraging results with respect to the previous year.