Contributions of Creativity to STEM Success During and After College: Understanding Core Cognitive Dimensions and Advancing Computational Approaches

Creativity is a fundamentally interesting and important aspect of cognition. It is already among the most valuable attributes of the STEM workforce, especially in engineering which requires generating and evaluating novel solutions to complex problems. It is projected to become even more vital in the innovation economy of the near future. Nonetheless, creativity is not generally integrated into engineering coursework nor is the explicit assessment of it a common part of student evaluation. Indeed, it has proven to be difficult to define and measure. More effectively integrating creativity into undergraduate engineering education requires knowing which dimensions of creativity contribute most to students’ future STEM success. In this project, researchers will parse creative cognition among engineering undergraduate students along core dimensions, including visual vs. verbal creativity and generative vs. evaluative creativity, and, in a longitudinal study, examine the relationships between creativity and engineering success during and after college. The substantial time, effort, and cost required for humans to score creativity limits the integration of such measurement into STEM classrooms and other educational and professional contexts. The researchers will use emerging computational approaches to transform the measurement of creativity, seeking to open new avenues for researchers to capture creativity in STEM on a large scale and in efficient and valid ways. Improved means of assessing creativity at such a scale may be especially important in light of the rise of hybrid creativity in which humans evaluate AI-generated ideas. Finally, traditional STEM assessment has been shown to lead to attrition among underrepresented minority and female students in STEM. Encouragingly, creativity is generally associated with less disparity by race and Socio-Economic Status (SES) than are traditional academic assessments, and generally does not show gender disparity. This work will lay the foundation for the eventual development of assessments that allow STEM educators to identify promising and creative students who might otherwise not persist in STEM, with downstream implications for broadening STEM graduate training and STEM workforce inclusion. The project has three primary objectives. In Aim 1 the research team will investigate how core dimensions of creative cognition (visual, verbal, generative, evaluative) relate to future STEM success in a 3-year longitudinal study. Using two cohorts of engineering students, one tracked during college and one tracked post-college, they will test verbal and visual measures of generative and evaluative creativity in both of these cohorts, including students’ evaluation of AI-generated ideas. Generative creativity will be scored by human raters. Evaluative creativity will be scored relative to human expert-rated criteria. These scores will be used to predict both college and post-college STEM success at a 3-year longitudinal timepoint. In Aim 2, researchers will test computational creativity scoring models and determine which best predicts human creativity ratings as well as which of those computational models best predicts future college and post-college STEM success, parsing the relative contributions of verbal and visual creativity. The human and computational creativity scoring approaches will also be examined to see whether they show less disparity for under-represented minority students than do traditional academic assessments. In Aim 3, the researchers will test computational creativity scoring approaches to predict college STEM success, using a large-scale data set of college applicants (n > 500,000) as a testbed. The dataset was compiled across 24 diverse U.S. universities (including public, private, and minority-serving institutions). The study will identify metrics that best predict human ratings of applicants’ creativity and use these metrics to predict STEM success (e.g., STEM Grade Point Average, STEM degree attainment). They will look to see whether disparity on race and SES is reduced relative to SAT scores. This project is supported by the EDU Core Research (ECR) program. ECR supports fundamental research that generates foundational knowledge that advances the research literatures in STEM learning and learning environments, broadening participation in STEM, and STEM workforce development.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.