The Truth About Affirmative Action: What the Data Reveals

Examining the data behind diversity and academic merit in college admissions.

While I now reside in the United States and have American citizenship, I’m only a naturalized scientist. The truth is that I was born and raised in a little coastal town a few miles north of Barcelona. 

Of course, I had to learn many cultural and logistical things about education and universities as I first worked as a university professor and later had children of my own. 

And among all things education, college admissions tend to puzzle me. 

Back where I was born being accepted into a public university is the most prestigious thing. Your high school GPA gets averaged with a universal selectivity exam that provides a final score for you. You use that score to apply to the universities of your choice. If a university has 200 spots to fill, the 200 students with the highest scores will be admitted. Period. 

On the other hand, you never pay more than $2,000/ year for tuition, and low-income students can attend for free and even get scholarships to help them pay for transportation, educational materials, and anything they may need. You need to keep good scores to keep these benefits.

I was one of those “low-income” students. Because of that, I was able to study for a bachelor’s and master’s degree fully paid and graduated debt-free. I then went on to earn a PhD in Australia. 

Just like me, there were many students from ethnic minorities who were able to get the same results I did. Admissions were based on qualifications, monetary help on income and inequalities. 

If you were to ask me, I think it was a fair system built to help everyone regardless of their background. However, I would need data to confirm that. 

Now let’s look at the United States. For a while, Affirmative Action helped students who would have had a harder time getting the opportunity of college education. It came about with the idea of helping minorities whose historical circumstances may prevent them from achieving their full potential. But now this protection is gone. 

I’m certain that, had I been born in the United States, my academic path may have looked very different. But did Affirmative Action actually help? Was its removal worth the cost it may have in society? Let’s see what science has to say about it. 

In the wake of the 2023 Supreme Court ruling in Students for Fair Admissions (SFFA) v. Harvard, colleges across the U.S. have had to rethink their admissions processes. Affirmative action, a policy aimed at fostering diversity by considering race in admissions, was deemed unconstitutional.

Advocates of the decision argued that it would create a fairer system by prioritizing academic merit over race. But is this the case? What does the data have to say about it?

A new study led by researchers at Cornell University paints an entirely different picture. According to their findings, removing race as a factor in admissions dramatically reduces diversity while offering little to no improvement in academic qualifications.

Ending Affirmative Action Harms Diversity Without Improving Academic Merit

But before you express your strong opinions on the topic, let’s look at how the study was performed. Shall we?

To explore the implications of race-neutral admissions, the research team trained an artificial intelligence-based ranking algorithm using four years of application data from a selective, engineering-focused university. 

The algorithm initially included all applicant information, such as grades, test scores, extracurriculars, and demographic details, including race. They then retrained the algorithm without race-related data and compared the composition of the top-ranked applicant pool under both scenarios.

The results provided a unique window into how policies could impact admissions outcomes without experimenting directly on real applicants, a critical consideration given the high stakes involved.

And what did the numbers have to say?

The shift to race-blind admissions led to a 62% drop in underrepresented minority (URM) students in the top-ranked applicant pool, from 53% to just 20%. 

Senior author Dr. René Kizilcec remarked, “We see no evidence that would support the narrative that Black and Hispanic applicants are admitted even though there are more qualified applicants in the pool.” In other words, when a BIPOC (Black Indigenous and People of Color) student gets accepted into a college, they have the merits necessary to be included. 

On the other hand, removing race did lead to a slight yet non-significant increase in average standardized test scores among top-ranked applicants; the difference was negligible. We’re talking about moving from a 1480 to a 1490 on the SAT (that’s a 0.68% improvement). 

This raises a pressing question: is less than a 1% gain in academic metrics worth the steep decline in diversity?

But let’s look at something else: the influence of algorithms and Arbitrariness.

Indeed, the study also covers another crucial aspect: arbitrariness in admissions. With so many qualified candidates applying, even small changes to an algorithm can drastically alter who gets ranked at the top. 

Co-first author Dr. Emma Harvey noted, “There were so many excellent applicants, the ranking changed substantially when the algorithm was trained with different random subsets of the data.”

In fact, when race data was removed, this arbitrariness worsened. Rankings became less consistent, meaning a student’s position in the top group was more likely to depend on the quirks of the algorithm rather than clear distinctions in merit. 

This undermines one of the central promises of race-neutral policies: fairness.

But let’s look at the big picture, shall we?

Critics of affirmative action often frame diversity and academic excellence as mutually exclusive. But this study suggests the opposite. The original algorithm (with race included) was able to identify a diverse group of students without sacrificing academic standards. 

The race-blind version, on the other hand, achieved neither diversity nor a meaningful boost in qualifications.

It’s also worth noting that the findings align closely with real-world data. When the study university published the demographics of its incoming class after the SFFA ruling, the decline in URM students mirrored the researchers’ predictions. 

However, the study’s authors acknowledged that not all institutions experienced the same outcomes. “Looking at how different schools responded to the SFFA decision will be very interesting for future research,” Dr. Harvey added.

This research highlights the unintended consequences of removing race from admissions considerations. While intended to level the playing field, race-blind policies may instead reinforce inequities by failing to address systemic barriers faced by URM applicants. 

It also highlights the importance of designing admissions algorithms responsibly, especially as more schools turn to artificial intelligence to process growing application volumes.

As Dr. Kizilcec aptly put it, “This work is critical to make sure that AI is used responsibly in admissions.” With thoughtful adjustments, it may be possible to create systems that promote both fairness and diversity without compromising academic standards. 

For now, though, the findings serve as a reminder that equity in education requires more than a neutral approach — it requires an informed one.