Affirmative Action Gone! What Happens Now?

To nobody’s surprise, the United States Supreme Court abolished affirmative action today. How does this affect college admission chances?

The two defendants in this case were Harvard College and the University of North Carolina (UNC). In this post, I will focus upon how this decision impacts Harvard and similar private colleges. A future post will focus upon selective public colleges like UNC.

My approach to estimating the impact is to use the internal models that Harvard had to disclose as part of this lawsuit. While the numbers below are specific to Harvard, the conclusions are applicable to all other private colleges with similar admission procedures. This includes all Ivy League colleges and many liberal arts colleges.

I will first outline my key conclusions, followed by the details of how I got there.

• “Unhooked”Applicants that do not benefit from "hooks". Examples of hoo... More admit rate improves only slightly: Even deep cuts in Under-Represented Minority (“URM”) admit rates would result only in a slight increase for unhooked students. As detailed below, Harvard’s admit rate for unhooked students increases from 2.3% to at most 3.2%. The reason this increase is so small is because the Court did not touch the admission hooks given to ALDC students (Athletes, Legacy, children of Donors, and Children of faculty). These students currently form about 30% of Harvard’s class, and will continue to be admitted at very high rates.
• URM rates won’t drop as much as expected: Colleges have expected this ruling for a while, and have prepared for it in ways I detail below.

How Harvard Used Race In Admissions

As part of the lawsuit disclosure, Harvard released a chart created by its Office of Institutional Research. Harvard had created four internal models estimating how the various hooks affected the racial composition of the class, and compared it to the actual demographics in the Class of 2017. Note that while the resulting chart does not have a category for URM, I will define it here as students that are Black, Hispanic, or Native American.

These models can be summarized as follows.

• Model 1: This model shows the expected student body if Harvard only considered academics (which consist of GPA, class rigor and test scores). Here URMs comprise 3.3% of the class.
• Model 2: This model adds consideration of legacy and athletics. Here URMs comprise about 4.8% of the class.
• Model 3: This model adds consideration of Harvard’s extracurricular and personal ratings. Here URMs would comprise about 6.8% of the class.
• Model 4: This model adds consideration of race and family income. Here the URM representation would be 22.1%.
• Actual: The actual URM representation for the Class of 2017 was 21.2%.

I should note that Harvard downplayed this research in its defense. However, given how well Model 4 fits with the actual racial composition of the class, I think it is valid for my estimate below.

Likely Effects on Unhooked Student Admissions

Based upon the above models, we can estimate that removing affirmative action could reduce URM representation by at most from 22.1% to 6.8% (Model 4 to Model 3). This is the worst-case scenario for URM students and represents a 69% decrease. For reasons I will explain later, I don’t think this 69% decrease is the realistic scenario today. But I will use it to calculate the maximum possible admit rate improvement for unhooked students.

For the Class of 2027, Harvard granted 1942 acceptances among 56,937 applicants, giving an admit rate of 3.4%, but that rate mixes both hooked and unhooked students. My previous analysis suggests that, for unhooked students, the real admit rate is closer to 2.3%, assuming URM students were still hooked.

Now let’s recalculate the result but without considering URM as a hook. As before, we start the analysis by backing out the applicants and acceptances from the ALDC students, giving us:

So once again, after taking out the ALDC applicants and acceptances, we know have 54090 applicants chasing 1359 remaining admit slots. URMs made up 29% of the Class of 2027, but if we cut that by 69%, they will instead make up about 9.0% of the class. For now we will still assume they made up 29% of the applicants (Harvard had a practice of recruiting URMs in order to equalize admit rates among groups). But after accounting for potential overlap between URM and ALDC, let’s use 25% of the applicants and 7.5% of the admittances.

So looking at the best case scenario from an unhooked student’s perspective, their admit rate could increase from 2.2% to about 3.2% (i.e. 1214 / 37756). In other words, it is still below the overall admit rate of the university as a whole, due to the remaining ALDC hooks.

Why URM Rates Will Drop Less Than Expected

I said earlier that I don’t think that Harvard will experience a 69% decrease in URM admit rates. There are three key reasons for this:

• Test-optional: Almost all elite universities required standardized tests prior to the pandemic in 2020. Everyone became test-optional starting in 2021. And while colleges like Georgia Tech and MIT now require tests, Harvard and its peers have remained test-optional. Since URM applicants as a whole score lower on standardized tests, their admission chances improve if they are not required.
• Use of Income: The Harvard disclosure data revealed that an applicant could get admission benefits from being URM, or low-income, but not benefit from both. Since many URM applicants are also low-income, their admission chances can improve from using low-income as an admission bump.
• Limited Use of Race: The Court ruling still allows the use of race in a limited way. Specifically, the Court ruled that “nothing in this opinion should be construed as prohibiting universities from considering an applicant’s discussion of how race affected his or her life, be it through discrimination, inspiration, or otherwise.”

There is no doubt that the ruling will decrease URM admissions; that is the entire reason why Harvard and UNC fought so hard against it. But because of reasons like the ones listed above, I do not expect more than a 30% URM decrease initially, and as colleges test the limits of “considering an applicant’s discussion of how race affected his or her life”, it might end up being only a 20% decrease.