Graduate schools around the United States are working to improve access to science, technology, engineering, and mathematics (STEM) in a manner that reflects local and national demographics. The admissions process has been the focus of examination, as it is a potential bottleneck for entry into STEM. Standardized tests are widely used as part of the decision-making process; thus, we examined the Graduate Record Examination (GRE) in two models of applicant review: metrics-based applicant review and holistic applicant review to understand whether it affected applicant demographics at The University of Texas MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences. We measured the relationship between GRE scores of doctoral applicants and admissions committee scores. Metrics-based review of applicants excluded twice the number of applicants who identified as a historically underrepresented minority compared with their peers. Efforts to implement holistic applicant review resulted in an unexpected result: the GRE could be used as a tool in a manner that did not reflect its reported bias. Applicant assessments in our holistic review process were independent of gender, racial, and citizenship status. Importantly, our recommendations provide a blueprint for institutions that want to implement a data-driven approach to assess applicants in a manner that uses the GRE as part of the review process.

No abstract

Graduate admissions committees throughout the United States examine both quantitative and qualitative data from applicants to make admissions determinations. A number of recent studies have examined the ability of commonly used quantitative metrics such as the GRE and undergraduate GPA to predict the likelihood of applicant success in graduate programs. We examined whether an admissions committee could predict applicant success at The University of Texas MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences based on quantitative metrics. We analyzed the predictive validity of admissions scores, undergraduate GPA, and the GRE for student success. We observed nuanced differences based on gender, ethnicity, race, and citizenship status. The scores assigned to applicants by the admissions committee could not predict time to degree in PhD students regardless of demographic group. Undergraduate GPA was correlated with time to degree in some instances. Interestingly, while GRE scores could predict time to degree, GRE percentile scores could predict both time to degree and PhD candidacy examination results. These findings suggest that there is a level of nuance that is required for interpretation of these quantitative metrics by admissions committees.