Engineers work within complex social-ecological-technological systems (SETSs), yet little is known about how engineering students conceptualize these systems during design. Capstone design courses, where students undertake authentic projects with external stakeholders, offer a valuable context for studying student mental models of SETSs. However, analyzing qualitative data about mental models at scale remains a significant methodological challenge, and sending sensitive interview data to commercial AI services raises concerns about participant privacy. In this paper, we describe an ongoing ECR-EDU Core Research project studying capstone design students’ mental models and present a novel methodological contribution: an analysis pipeline using locally hosted, open-weight large language models (LLMs) to extract entities from qualitative data, score them along social, ecological, and technological dimensions, and compare across groups. We validated this pipeline using synthetic data designed to emphasize different SETS dimensions, and the pipeline successfully distinguished between groups. Additionally, we report preliminary observations from interviews with 28 capstone design students. These methods and initial findings have implications for researchers studying engineering design cognition and for instructors seeking to understand how students consider broader system impacts in their design work.
http://orcid.org/0000-0002-2202-6928
Virginia Polytechnic Institute and State University
[biography]
http://orcid.org/0000-0002-4255-3266
Virginia Polytechnic Institute and State University
[biography]
http://orcid.org/0000-0001-8837-5226
Virginia Polytechnic Institute and State University
[biography]
The full paper will be available to logged in and registered conference attendees once the conference starts on June 21, 2026, and to all visitors after the conference ends on June 24, 2026