Tracking and Supporting Newcomer Well-being in Science, Technology, Engineering and Mathematics

Being new to STEM can be stressful. While newcomer stress can resolve itself over time, it can also foreshadow issues like a lack of a sense of belonging, loss of interest, and ultimately intent to leave the field. While human instructors are seen as responsible for creating caring cultures in the classroom, this is not always feasible in higher education, as they must often deal with large class sizes, competing responsibilities, and a lack of indicators to signal and intervene on the student experience. Adding urgency to the matter, prior work suggests that silent struggles can be more common and detrimental to race and gender minorities in STEM. How might we scalably track and support emotional struggles in STEM classrooms? Can we support affect with targeted use of data, complementing its current use for pushing performance and accountability? My dissertation responds to these questions through a mix of experimental and design research that focuses on novice affect in STEM. Chapters one and two introduce two separate efforts to detect latent student affect for new STEM learners. Compared to study 1, which takes place in a 1:1 laboratory setting, study 2 moves detection efforts to a more ecological scenario of a group workshop, yielding lower but promising levels of performance. Study 2 additionally collects frequent stress reports through ecological momentary assessment (EMA), which is used to explore stress trends across demographic groups, as well as its impact on learning and motivation. In chapter 3, the findings from a design-based research study that tested AI-augmented periodic feedback in a makerspace course are discussed. Results show clear potential for automatic affect and motivation support, particularly through the mediating variable of classroom climate.