Use of the GRE in Graduate Admissions

May 20, 2019

Dear Colleagues,

Graduate admissions processes should identify candidates who are likely to succeed in our programs and do so in a way that is fair and equitable. In recent months, several campus groups, including the Equity, Inclusion and Anti-Discrimination Advocacy Committee of the University Senate, the University the Council on Graduate Studies, and the attendees at the graduate studies retreat have discussed the use of GRE scores in graduate admissions. I am particularly interested in the extent to which the use of these scores promotes or hinders the goals of graduate admissions, including the goal of promoting an inclusive and diverse environment. These discussions have been motivated by several factors, including:

  • Studies showing that GRE scores are biased against women and students from under-represented groups.
  • Studies concluding that scores are not predictive of key measures of success in graduate school.
  • The financial burden of paying to take the test, which may disproportionately affect low-income students. For example:  
    • The general test in the U.S. costs $207 and sending scores to additional programs costs $27/institution.
    • GRE test prep courses cost between $700 and $2,500 (Kaplan). Students taking prep courses tend to score better on these tests, creating another barrier for fair assessment of students of different economic means.
  • GRE scores have been eliminated as a reporting requirement for some federal agencies.
  • Eliminating the GRE requirement has resulted in increased numbers of applications, especially applications from students in underrepresented groups.

[See references below for additional information.]

Similar observations can be made about other standardized tests used for graduate admissions, but we have more flexibility around our use of the GRE compared to other standardized tests. In fact, several Pitt programs, including the Biological Sciences PhD program in Arts and Sciences (in 2017) and all PhD programs in the School of Medicine (in 2018), have elected to eliminate the GRE as an element of their admissions processes. Effects of these changes on applicant quality, diversity, and success will be evaluated in the coming years.

Personally, I believe that the evidence of gender and racial bias in the GRE is strong and that evidence of their predictive value for PhD student success is weak. Pitt has historically allowed a high degree of autonomy in graduate admissions and I am not proposing to change that autonomy or make any change in university policy at this time. Some programs are required to compile GRE or other standardized test data by their accreditation groups, and the University will not interfere in that process. However, given the importance of this issue, I am writing to clarify policy and make recommendations about the use of the GRE by graduate and professional programs across the University.

  1. There is not a University requirement that graduate programs use the GRE or any standardized test score as part of their admissions processes. Decisions about requiring tests and use of test scores are at the discretion of the schools and/or programs.
  2. If a program elects to use the GRE, members of their admissions committees should be aware of the guidelines that ETS, the creators of the GRE, provide. These guidelines include a recommendation against using “cutoff” scores. As described by the GRE:

“Using a minimum GRE score as the only criterion for denial or acceptance for admission or a fellowship award is not good practice because it overinflates the role of one measure of an applicant's value over others.”

  1. Programs should engage their faculty in a discussion of how and why they use GRE scores in their admissions processes and how they could assess the value that they gain from requiring GRE scores. In some situations, GRE scores may strongly indicate educational achievements and/or abilities that are important predictors of student success. If so, it should be possible to demonstrate this using data on GRE scores and outcome measures in that program. The programs should consider adopting best practices (such as holistic review) and developing more data-driven approaches to admissions.
  2. Programs requiring the GRE should strongly consider processes by which this requirement could be waived for students for whom taking and reporting the test would represent a financial hardship.

My perspective is that graduate admissions decisions are difficult, with many factors to consider and very little standardized information on which to base a decision. Therefore, my initial thought was that eliminating an objective and standardized measure would reduce our ability to identify the best students for our programs. However, the more that I evaluate the evidence, the more concerns I have about what the GRE measures and predicts, which in turn heightens my concerns about the equity of requiring a test for which performance is clearly correlated with income, race, and gender. Use of the GRE thus may be at odds with the goals of the University of Pittsburgh strategic plan, which include attracting a diverse population of students.

Conversely, we should also not ignore the fact that occasionally standardized tests provide an opportunity to discover talented students who might not perform well on other metrics, as in the story of the recent Nobel Prize winner who believes that she got into college despite poor grades because of near perfect SATs. 

If you have questions or comments, feel free to contact me. I anticipate that this will be a topic of discussion in future UCGS meetings and other venues in which graduate education is discussed. If your department or program would like assistance in changing policies around graduate admissions or in communicating to potential applicants about these changes, I am happy to set up a time to discuss this issue.