Bridging Conventional Admissions Metrics and Undergraduate Engineering Student Non-Cognitive and Affective Factors.

Undergraduate admissions serves as a gatekeeper in higher education and often includes measures of secondary school academic performance, including standardized exams (e.g., SAT, ACT, AP, IB) and grade point average, which have weak to modest correlations with undergraduate academic performance. Additionally, these measures have been critiqued as biased and prioritize students from privileged backgrounds. More recently, non-cognitive and affective (NCA) factors—psychosocial constructs spanning behaviors, feelings, and emotions—of accepted students have emerged as stronger predictors of future academic performance compared to conventional admissions data. NCA factors also provide a multifaceted understanding of students that can direct actions for addressing representation and inclusion issues in engineering. In this paper, we conducted a $k-\text{means}$ clustering of admissions metrics and NCA factors (i.e., belonging, identity, the Big Five) to understand if there are patterns among these variables within a large sample of engineering undergraduate students from two national surveys $(\boldsymbol{n} =\mathbf{6,050})$ . Second, we evaluated the differences between students of varying backgrounds (i.e., gender, race/ethnicity, first-generation status) on these metrics and by cluster. Finally, we assessed the influence of the type of institution (e.g., public or private, doctoral degree granting, tuition guarantee, etc.) by cluster. Understanding how commonly-used admissions metrics relate to NCA attributes may reveal whether such a relationship is appropriate to consider in a holistic admissions review process that facilitates the matriculation of a more diverse engineering study body. Our results indicate that readily-available admissions data signifies the need for tailored support for matriculated students in order to support persistence of the incoming students through to an engineering degree.