Can a liberal society do affirmative action right?

The quota trap.

California State University students protest former Ku Klux Klan leader David Duke as he speaks in favor of abolishing affirmative action in the state of California. (Kim Kulish/Getty Images)

In 1996, Californians voted 55 to 45 for Proposition 209, which, among other things, barred public colleges and universities from considering race or ethnicity in admissions.

Twenty-four years later, Californians voted on whether to repeal the proposition. The state had become far more liberal since 1996, so the organizations that led this campaign had reason to think they might succeed: Bill Clinton had earned 51% of the California vote when 209 passed; Joe Biden took 63.5% in 2020. It had been more than a decade (it has now been nearly two) since a Republican won statewide office.

But the vote on repealing Proposition 209 was even more decisive than the original one. By a margin of 57 to 43, Californians again voted against permitting public institutions to discriminate or grant preferential treatment on the basis of race, sex, color, ethnicity, or national origin. The proposition wasn’t even overwhelmingly popular with the populations it purported to benefit: while Black Americans did support it in polling, some polling suggests that Latino voters opposed it by sizable margins (30% in favor, 41% against).

I am frequently a critic of the California electorate, but I think that the state’s voters got this right in 1996 and again in 2020.

America can do better than race-based affirmative action

It is completely reasonable for admissions officers, when evaluating college applications, to take into account the disadvantages that prospective students have encountered.

Sure, it’s both cringe and undesirable to incentivize students from comfortable backgrounds to desperately cast around for a way to frame their lives as tragic. But it’s straightforwardly a good thing to take into account whether a student had to pay their own bills or had a private tutor, whether they were encouraged by teachers or discouraged, and whether their parents knew the rules of the game — and racism is one form of disadvantage students might encounter and overcome.

I think that students learn better on a campus where they are presented with views, perspectives, and experiences that are unfamiliar to them. I think it expands their horizons and makes them wiser, more thoughtful, and more compassionate.

The problem is that, as implemented, affirmative action programs don’t actually do this.

In practice, race is not just treated as one possible way in which a student might be disadvantaged, alongside others like parental education or the stability of their housing situation. Instead, it is consistently treated as more important than any of those.

When the Harvard admissions committee had slightly more “tentative yes” students than they wanted to extend offers to, applicants were discussed from a list with only four pieces of information: “legacy status, recruited athlete status, financial aid eligibility, and race.”

Most measures of experienced, day-to-day racism find there is anti-Asian racism in the United States, often of similar magnitude to anti-Hispanic racism. And yet, because Asian students are often overrepresented at universities relative to the U.S. population as a whole, affirmative action programs make it much harder for an Asian student than a comparably talented white student to be admitted to college. And if you look at more granular research, South Asian students are strongly disadvantaged relative to even East Asian students with the same scores, yet by most metrics, there is more anti-South Asian than anti-East Asian racism in America.

During the discovery process for Students for Fair Admissions v. Harvard, the 2023 Supreme Court decision that struck down affirmative action, the public got a glimpse at how, mechanically, the process actually played out in elite admissions offices.

Reading some of the back and forth between admissions officers is disillusioning for those of us who hoped for a sensitive and thoughtful process that recognized the nuances of student identity and the fact that students can be impacted by prejudice in a wide range of ways.

“Brown?!” one UNC admissions officer wrote in an admissions group chat about a specific applicant. “Heck no. Asian,” another replied.

“If it[’]s brown and above a 1300 [SAT,] put them in for [the] merit/Excel [scholarship],” another admissions official wrote.

These interactions make no sense, of course, if you think of the purpose of affirmative action as being either redress for past wrongs or as being about the perspectives a candidate would bring to campus.

But they make more sense if you understand these programs as an effort to achieve specific percentages of admitted students from the specific target demographics that colleges report publicly. And this effort has now subsumed efforts to systematically help disadvantaged students, and it has done so in a manner that, frankly, seems out of place in a liberal society.

Quotas are not a good vision of pluralistic liberalism

It’s true that many countries are trying to make a multicultural or a multiracial society work using racial quotas or a race-based affirmative action program. China has historically awarded extra points on university examinations to racial minorities. Singapore mandates quotas for housing, and Brazil has quotas for higher education and the civil service. Canada requires federally regulated large employers to achieve race and sex equity through affirmative action hiring programs. South Africa and India both have affirmative action programs.

But from the perspective of a political liberal, this is obviously challenging.

Our foundational commitment is that every person is an individual, created equal, and deserving of equal treatment under the law and equal opportunity. That means you shouldn’t get judged by the average qualities of your group. You shouldn’t be judged as less impressive because other East Asians have, on average, higher test scores, and you shouldn’t be judged as more impressive because other Hispanics have, on average, lower test scores.

For all its flaws, America is a more racially integrated society than most of the nations I mentioned above. In measures of employment discrimination, America is consistently among the countries where the employment effects of a foreign or racial-minority name are the smallest. Approval of interracial marriage is much higher in the U.S. than in Europe or Asia, and rates of intermarriage are correspondingly higher as well. Ten percent of Americans identified as multiple races on the 2020 Census, and 15% of Americans under 18; the share of the country that can’t fit into neat racial boxes will only continue to grow.

When affirmative action programs largely began in the 1960s, the aim was to try to redress the harm that centuries of discrimination had done to Black Americans by giving them the leg up into American institutions and halls of power, which some white Americans enjoyed because their ancestors had.

I think that was appropriate. I also don’t think it bears much resemblance to how admissions offices actually evaluate candidates today. In practice, affirmative action programs generally amount to picking some cultures of origin and declaring that they should benefit while declaring that others should be ineligible, and the way we pick those groups is increasingly incoherent.

Under our existing system, a student whose family descent is Filipino might be rejected from college admissions because there are too many students from India and China with great scores. Whether a multiracial student is admitted may turn ultimately on which races they disclose, and there are a bunch of low-quality polls suggesting very high rates of lying about race on college applications.

Which brings us back to our admissions officers deciding whether to refer students for a scholarship or not based purely on whether they’re “brown.” In any workplace in America, this would be illegal under the Civil Rights Act, which wisely barred discrimination on the basis of race in employment.

Those text messages aren’t defensible under any of the idealistic accounts sometimes given of affirmative action: that we want a diverse range of students who can learn from each other , or that we want to be mindful of ways a student encountered hardship .

Colleges, of course, have an interest in students from a wide range of backgrounds. But what we see there isn’t an interest in the complicated and varied backgrounds of potential students. It is a reductive obsession over skin color—an assumption that once you know someone’s broad racial category, you know their value.

The American people love race-blind policy to help the disadvantaged

While my main point here is not about popularity, it’s worth acknowledging that affirmative action is deeply unpopular.

A plurality of Black Americans agreed with the Supreme Court’s decision to overturn race-based affirmative action, with 47% saying colleges should not be allowed to consider race and only 36% saying they should. Hispanic Americans agreed with the Supreme Court by even larger margins: Fifty-five percent said colleges should not be allowed to consider race, and only 28% said that they should.

This issue is underwater even in the populations that it purportedly benefits.

By contrast, polls generally find support for affirmative action on the basis of class, benefitting low-income students at the expense of rich ones.

Even about half of rich people said they feel that poor students should be advantaged in admissions over their own kids. Americans are broadly a generous people who admire students who make good from disadvantaged backgrounds, and they are supportive of programs to enable that.

One study argued that Americans support DEI programs when the pitch is that “the goal is to reduce barriers faced by disadvantaged groups, in order to create a better merit-based system.”

The researcher’s takeaway was that DEI programs needed to be pitched differently, but I think that’s mistaken: They need to actually be different.

You can’t convince people that “refer this student for a merit scholarship if they’re brown, but not if they’re Asian” is creating a better merit-based system. But you can absolutely design a real merit-based system, and when you do, you’ll find broad public support for it.

I should acknowledge that income-based programs do not produce the same results as race-based programs.

If you don’t use race at all in evaluating candidates, but instead invest aggressively in the proxies for economic disadvantage — low-income families, underfunded schools, first-generation college students — you do see a decrease in Black and Hispanic attendance at top schools.

Research on the UC system after California banned race-conscious admissions has suggested that schools get about 60% to 70% as many Black and Hispanic students when they don’t explicitly consider race. Work from the UC system also showed that some of those students will be worse off.

These numbers may be a floor rather than a ceiling, though. Recent research on the effects of the 2023 Supreme Court decision banning the use of race in admissions suggests much smaller drops in Black and Hispanic student enrollment at highly selective colleges than the UC system warned of — just 18%, about half of what was feared.

The difference could be explained by the changes happening across the board rather than in a single state: if a single state bars considering race in admissions, some of its student pool may flow elsewhere, while if the practice is banned nationwide, the “pool” is effectively fixed, and the effect smaller.

The long shadow of Joe Biden

We’re at war, gas prices are through the roof, school test scores keep falling, and a terrifyingly large share of young people openly endorse various forms of religious hatred while a shrinking share value free speech. This is yesterday’s fight, and rendered moot since the Supreme Court already ruled that the use of race in admissions is not allowed. Why am I still talking about this?

Until recently, I thought the issue was settled, but a recent study changed my mind.

Political scientists David Broockman and Josh Kalla examined which issue positions, attributed to a candidate, made them more or less popular with voters. Their big-picture takeaway is that we should stop talking about “moderation” — voters don’t look for candidates whose views are in some sense objectively “moderate.”

But voters do care, intensely, about a candidate’s policy positions, and about whether a candidate shares the voter’s views.

And the single issue that made the biggest difference to voters — the one where a candidate adopting the “elite middle” position, rather than the progressive one, dramatically increased participants’ willingness to vote for them — was affirmative action.

When a Democratic candidate moved from the party’s standard position on affirmative action (“universities may consider race in admissions to increase diversity”) to the still-fairly progressive position, “universities may not consider race directly, but may consider essays or other materials showing how applicants were affected by racism,” they gained 4.5 points of vote share.

That was, by a significant margin, a bigger deal than any other issue they studied except for one: Replacing racial preferences in small business loans with preferences for low-income applicants produced a boost of the same magnitude.

Why did these two issues—generally not very high-salience ones—carry so much weight in voters’ view of candidates? Certainly, part of the reason is that in this task, voters are given only a little bit of information about a candidate, so low-salience issues can carry the day. These issues are not top of mind for voters in ordinary elections. But still, the effect size is striking.

The study suggested that “colleges can consider a candidate’s experience of racism, but not their race” — still a progressive position! — does more to win cross-pressured voters than restricting asylum-seekers.

I spent a while puzzling this over. I built a copy of the task from the paper and asked a bunch of acquaintances to take it. I looked at a lot more quantitative and qualitative research on voter views. I have a theory of my own now, though it’s speculative:

I think that Democratic candidates benefit enormously when they can credibly signal to voters that they’re not Joe Biden. And while on most issues, breaking with Biden means satisfying some voters and losing others, on this issue, it just straightforwardly grows the tent.

Joe Biden left office deeply unpopular, more unpopular than any president but Donald Trump. Independents in particular are very sour on him, and voters who hated his presidency tend to still be extremely negative on the Democratic party even as they sour on Donald Trump.

They don’t, to be clear, mostly hate Joe Biden over college admissions policies: polls have consistently found that they mostly turned on him over immigration and inflation. But — running in 2020, as the country grappled with George Floyd’s death and the demands of the Black Lives Matter movement — Biden embraced a very directly race-oriented view of Democratic politics. That included promising that he would make key appointments on the basis of race, most famously promising to appoint a Black woman to the Supreme Court, a promise that was in fact unpopular even with Democrats (76% of Americans said they wanted him to consider all possible nominees). He sharply condemned the Supreme Court ruling against race-based admissions.

Moves to moderate on policy, if they’re substantive at all, involve adopting specific positions that many Democrats dislike. Move to the center on immigration, and you anger some voters while satisfying others. This is, Broockman and Kalla found, a crucial effect to understand if you want to argue over moderation: “Candidates moving towards the other party win some voters but lose others who preferred their party’s position.”

That makes “do not directly use racial preferences in public policy” unique among the issues tested — strongly associated with the establishment and with the unpopular Biden administration, yet without any large constituency in favor (again, Black and Latino voters agreed with the Supreme Court!). Opponents also feel much more strongly about it than supporters do, so that there’s almost no one you lose by articulating a vision for race-blind admissions.

The paper can’t settle why voters feel the way they do. And to be clear, it would not in itself constitute good reason to adopt this position. I don’t, in fact, want candidates who campaign with their nose to the polls.

I think that candidates should adopt this position because it is right, not because it is popular: the more liberal, more individualist, position, better on the merits for a modern multiethnic democracy.

But it does seem to be popular, and it may be a way for candidates to carve out a brand distinct from the deeply unpopular Biden one. I think its significance to voters is at least sufficient reason to discuss the issue on its substantive merits.

More from The Argument: