After doctors graduate from medical school, they enter a residency program, applying through the Electronic Residency Application Service (ERAS) run by the Association of American Medical Colleges (AAMC).
The National Resident Matching Program (the NRMP), a private organization, uses an algorithm to place applicants into residency and fellowship positions.
Beginning in 2022, the NRMP began collecting demographic data on applicants who used its service. This data collection is voluntary.
However, this decision was explicitly motivated by the NRMP’s goal to address diversity in residency programs.
“The decision was driven by support from national learner organizations and members of the broader medical education community who viewed the NRMP as the entity best positioned to lead efforts to characterize the current state of diversity in the transition to residency and encourage greater equity in the ranking and matching processes,” the NRMP stated in a 2023 research brief addressing its demographic data collection.
“When registering for the Match, applicants are invited to provide information on characteristics including sex and gender, race, and ethnicity as well as socio-economic status, first-generation education, and disability,” the NRMP continued.
While the NRMP makes clear that its demographic data “will never be incorporated in any way into the matching algorithm,” the NRMP nevertheless uses its findings to advocate for diversity initiatives, some of which appear to be racially discriminatory.
“There is a clear need to build greater applicant diversity earlier in the pipeline so as to eliminate the imbalances in representation, race in particular, that drive findings like these,” the NRMP stated in its research brief.
Indeed, the NRMP explicitly cites diversity initiatives that “have focused on modifying selection, interview, and ranking processes for residency” to increase the representation of “URiM” students (students from underrepresented minority groups).
That sounds an awful lot like racial discrimination.
One such referenced initiative explicitly devalued the role of applicants’ test scores for an emergency medicine residency program at the Emory University School of Medicine, reasoning that “racial disparities exist in standardized tests.”
Another initiative explicitly prioritized URiM applicants in the interview process at the University of Utah Health.
And still another included specific recommendations for program admissions officials to favor racial minorities in multiple stages of the application process.
Moreover, although the NRMP may not be explicitly using race to match applicants to residency programs, the organization outright admits that its data collection efforts are to achieve “greater diversity and equity in medicine.” An excerpt from its 2022 annual report reads as follows:
There is much discussion about the need for greater diversity and equity in medicine, but to achieve that objective, the origins of underrepresented in medicine must be examined. For the NRMP, that means revealing and analyzing the applicant profile, not just along racial and ethnical lines but also gender identification, socioeconomic status, and disability. It will benefit the profession to understand how different demographic characteristics are viewed, integrated into the transition to residency process, and impact outcomes.
Another excerpt states the NRMP is intent on “leveraging applicant demographic and specialty preference data to address workforce equity, especially for underserved populations.”
In less-Orwellian terms, the NRMP is making clear that its demographic data collection efforts will help residency programs better promote “diversity” (read: engage in racial discrimination).
Imagine how applicants of disfavored racial groups feel: they are trusting an organization to place them into a program that will further their career, in which they’ve invested a nearly-unfathomable amount of time and effort.
And that organization is enabling discrimination against them on the basis of their race!
That is unconscionable.


