Author(s):
A barrier to improving teaching of complex skills in science is the ability to measure them at larger scales. We have developed a digital, performance-based assessment of graph construction competence that can be (mostly) auto-scored and used at large scale. We call the assessment GraphSmarts and have developed six storylines in partnership with instructors from diverse institutions to provide different real-world biological contexts in which to graph. In GraphSmarts students test predictions through graphing and answer some closed and free response questions which allows us to gain insight into their graphing practices, including reasoning. Our guiding research questions are: What are the graphing practices of undergraduate biology students from different institutions and course contexts? In what ways does biological context impact students’ graphing practices? We are in our final semester of student data collection and thus far ~3,400 students from diverse institutional contexts have completed our assessments as an activity in their biology class. We have developed autoscoring algorithms for students’ graphs and responses to intermediate constraint question types and are finalizing the development of natural language processing models to code free-response data. Our data will provide insight into which graphing practices students do well and which students need support in developing. We plan to collaborate with instructors to leverage our insights together with instructor knowledge based on their teaching experience to build graphing instructional tools and an assessment plan to use in their specific classrooms that are relevant to their student populations and institutional context.
Coauthors
Eli Meir, Ryan Baker (Co-PIs not attending the meeting)
