Bjarte Hannisdal, Einar Iversen
Data analysis and statistics play a key role in the geosciences, but have been nearly absent in traditional geology BSc curricula. At our department, geology students have historically been offered a one-week intensive lecture-based course at the MSc level. In 2017, the authors launched a major revision of both form and content of this course. Our goal is for students to adopt computational practices as a means of developing their expertise in solving authentic, ill-structured problems (Scherer et al., 2017, J. Geosci. Educ. 65).
As a first step we reoriented the form of instruction towards real-time problem-solving using the programming language R and the RStudio desktop interface. Both instructors were present during the organized instruction, one demonstrating computational practices and the other demonstrating problems on the blackboard. This paired instruction enabled continuous peer review and feedback on the form and content of the course. Students performed all computations on their own laptops, and also engaged in group activities (such as rolling toy dice in the corridors to experience the central limit theorem). Assessment was based mainly on an inquiry-based term project designed to let students define and test statistical hypotheses in R using their own data or other published data relevant to their research topic. In a course evaluation group interview, students noted that they would have preferred to work with real data from the start.
We intend to further develop and test new learning activities in a revised course offered in 2019. Our primary hypothesis is that data practices and computational practices (Weintrop et al. 2016, J. Sci. Educ. Technol. 25) significantly improve student learning in the context of authentic, ill-structured problem-solving (Holder et al., 2017, J. Geosci. Educ. 65). To test this hypothesis, we will assign students to an experimental group that uses computational practices, and a control group that reads the same instructional material, and use pre- and post-instruction interviews to assess their progress from novice towards expert-like thinking.
We solicit input from ISSoTL18 participants on our proposed experiment, specifically on setting up integrated assessment and evaluation of computational practices. A challenging “threshold” concept in elementary statistics is the central limit theorem (CLT). With a computer, however, students can discover the CLT themselves without any prior theoretical knowledge. In our presentation, we invite ISSoTL18 participants to also make this discovery by playing with virtual dice using simple computer code.
