Racial preferences at elite American universities

Last Updated on December 1, 2021

In this post, I explore racial preferences at elite universities in the United States. I start by outlining the basic racial demographics of top universities. Then I compare this data to the racial demographics of American high schoolers with the highest level of academic achievement. This comparison highlights the consequences of extreme racial preferences in the admissions process of our elite institutions of higher education. Next, I estimate the demography of top universities after separating Jewish and non-Jewish whites. Surprisingly, these estimates suggest that non-Jewish whites are perhaps the most under-represented racial group at these elite institutions.

Following that, I review studies that attempt to quantify the magnitude and patterns of racial preferences at selective universities. These studies show that elite universities have little to no preferences for low-SES or low-income applicants. Next, I briefly cite a few reviews of affirmative action case law in order to understand the primary legal defenses of racial preferences that have been presented by selective universities in landmark Supreme Court cases. The review of affirmative action case law leaves me with a few important questions that I believe have not been adequately addressed by elite universities. Finally, I end the post by explaining why racial preferences at elite universities are important.

Demographics of elite universities

Before presenting the demographics of students at elite universities, we should first understand the racial demographics of the college-aged portion of the general population so that we have a baseline to use for comparison. Recent census data [archived] hosted by the National Center for Education Statistics (NCES) contains estimates of the population disaggregated by different age groups. One of the age groups includes individuals between the ages of 18 to 24 years old. I’ll use this age group as an estimate of the college-aged population. This data shows that, as of 2020, the percentage of the population within this age range by race/ethnicity was as follows:

  • White (52.9%)
  • Hispanic (23.1%)
  • Black (14.0%)
  • Asian (5.7%)
  • Multi-racial (3.3%)
  • American Indian / Alaska Native (0.8%)
  • Pacific Islander (0.2%)

Demographics at each school

For the purposes of this post, an “elite” university is defined as a top 20 national university according to the 2022 U.S. News & World Report rankings. The universities that make the list are not surprising. The top 5 universities (in order) are Princeton, Columbia, Harvard, MIT, and Yale. All 8 Ivies made the top 20, with the lowest-ranking Ivy (Cornell) ranked at #17.

To get information on the racial demographics at each university, I used data from the Integrated Postsecondary Education Data System (IPEDS), the core postsecondary data collection program for the NCES. This program provides annually updated data on the racial demographics of the undergraduate student body for every (?) major university in the country. This source is superior to statistics hosted on official university pages for a number of reasons. Firstly, many universities fail to include the percentage of students who are white, even though they include the statistics for other racial/ethnic groups in their statistics. For example, here are the racial/ethnic groups posted on Harvard’s page [archived] for admissions statistics:

Same thing for the University of Chicago’s page [archived] on the accepted class of 2025:

Secondly, many other universities posted statistics for white students, but included Hispanics under “white” and did not have a separate multi-racial category, resulting in totals that exceeded 100%. For example, here are the diversity statistics posted on Yale‘s page [archived]:

Same thing for Columbia [archived]:

For example, while Yale’s official webpage reports that 50% of non-international students are white, the data reported to the IPEDS shows that only about 39% of non-international students are white (35%/90% since 90% of students are domestic), once we exclude Hispanic students and multi-racial students.

The IPEDS data is preferable since it uses consistent classifications across each university, and it is consistent with the classifications used by official Census data. Furthermore, I believe the IPEDS classification better captures important social differences between ethnic groups. Non-Hispanic whites have very different social realities than Hispanics and multi-racial individuals, so it’s best to separate them in the statistics. For that reason, “white” will always refer to non-Hispanic, non-mixed whites in this post, unless stated otherwise.

That being said, here are the racial demographics at the top 20 American universities (Ivy Leagues are in bold):

Johns Hopkins28%26%16%8%7%0%3%12%
Notre Dame68%5%11%3%6%0%1%6%
  • “Other” includes subjects classified as “American Indian or Alaskan Native” or “Native Hawaiian or other Pacific Islander”.
  • “NRA” stands for non-resident alien.

Summarizing the data

Now, I will summarize the data reported in the above table. I will exclude non-resident aliens and students with a race/ethnicity categorized as “unknown” or “other” to focus on the major racial groups of domestic students — white, Asian, Hispanic, black, and multi-racial students. For these groups, I calculated the following two figures for both the top 20 (T20) universities and Ivy Leagues:

  • The weighted average of the percentage of students by race/ethnicity was calculated using the data presented above and data on the total enrollment size from the NCES links.
  • The enrollment rate ratios at elite universities between non-whites and whites were calculated using these weighted averages and the demographic data on the general population. For example, the data below shows that Asians are about (21.8%/5.7%)/(36.6%/52.9%) = 5.53 times as likely as whites to enroll in a top 20 university. See the following bullet points for more on why this arithmetic is a valid way to determine relative enrollment rates:
    • The relative enrollment rate calculated in the above example (5.53) is based on evaluating the expression (21.8%/5.7%)/(36.6%/52.9%).
    • Note that this is [(% of Asians enrolled) / (% of Asians in gen pop)] divided by [(% of whites enrolled) / (% of whites in gen pop)].
    • This evaluates to [(# of Asians enrolled) / (total # enrolled)] / [(# of Asians in gen pop) / (total # in gen pop)] divided by [(# of whites enrolled) / (total # enrolled)] / [(# of whites in gen pop) / (total # in gen pop)].
    • This evaluates to [(# of Asians enrolled) / (# of Asians in gen pop)] divided by [(# of whites enrolled) / (# of whites in gen pop)].
    • This is equivalent to the enrollment rate for Asians divided by the enrollment rate for whites (where “enrollment” rate indicates the number of students enrolled per person in the college-aged population).

All of these calculations (with links to relevant sources) are published in a Google spreadsheet available here. These calculations for each race/ethnicity are available in the following table:

Weighted Average (T20)36.6%21.8%14.7%6.5%6.1%
Weighted Average (Ivy only)37.3%20.2%12.6%7.7%5.6%
General Population (18 – 24 yo)52.9%5.7%23.1%14.0%3.3%
Enrollment Rate / White Enrollment Rate (T20)1.005.530.920.672.66
Enrollment Rate / White Enrollment Rate (Ivy)

Some interesting findings regarding the white population stand out from this data:

  • White students constitute under 50% of the undergraduate student body in 19 of the top 20 universities.
  • White students constitute under 40% of the undergraduate student body in 9 of the top 10 universities.
  • The weighted average percentage of students who are white (37%) in these elite universities was surprisingly low to me. I imagine this will also be surprising to those who treat many of these universities as “predominantly white institutions.”

There are also other interesting findings to be found when comparing different racial groups:

  • Every racial group except for Asians and multi-racial students are under-represented.
  • The T20 enrollment rate for Asians (share of enrolled divided by share of the general population) is about 5-6 times the same rate for whites. This data is not too surprising given superior Asian academic performance in secondary school (see below).
  • Surprisingly, the enrollment rates for blacks and Hispanics was not much lower than the rate for whites. This is surprising because there are massive differences in academic achievement between these groups in secondary school (see below). Despite large gaps in academic preparation, the enrollment rate at T20 universities for Hispanics and blacks were only 8% and 33%, respectively, lower than the same rate for whites. Among Ivy League schools, the enrollment rates for both Hispanics and blacks were only 22% lower than the same rate for whites.

Racial differences in academic achievement

In order to put the magnitude of the enrollment disparities in perspective, it will be helpful to compare them to the racial disparities in academic achievement in secondary school. I’ll do this by looking at racial differences in scores of two different standardized tests:

  • National Assessment of Educational Progress (NAEP) scores: the NAEP conducts the largest nationally representative assessment of academic skills among students from 4th grade to 12th grade. For my purposes, I will use data on mathematics and reading scores in the 12th grade.
  • SAP scores: the SAT is the most common standardized test submitted to universities for admission. Most students at these elite universities submitted SAT scores during the admissions process.

NAEP scores

The NAEP data shows large racial gaps in mathematics and reading achievement. These gaps have remained fairly stable for over 20 years. Here are the 12th grade mathematics and reading scores by race:

  • Reading [archived]: In 2015 (the most recent year available), the mean reading scale score for whites was about 295 points. Blacks and Hispanics performed significantly worse, scoring just 266 and 276 points, respectively, about 20-30 points lower than whites. Asians performed about the same as whites, scoring 297 points, just 2 points higher. In 1992 (the earliest year available), the white-black and the white-Hispanic was gap was 24 points and 19 points, respectively, which is actually lower than the gap today. So the gap has not diminished significantly in over 2 decades.
  • Mathematics [archived]: In 2015, the mean mathematics scale score for whites was about 160 points. Blacks and Hispanics performed significantly worse, scoring just 130 and 139 points, respectively, about 20-30 points lower than whites. The Asian advantage in mathematics was more pronounced than the Asian advantage in reading, with Asians having a mean score of 171 points, about 11 points higher than whites. Unfortunately, the earliest data for 12th grade mathematics scores only goes back to 2005 when the white-black gap (31 points) and white-Hispanic gap (24 points) were similar in magnitude to the gaps today. Fortunately, a different mathematics scale has been used in alternating years with data [archived] from 1973 to 2012. This shows that the black-white and Hispanic-white gap has not narrowed significantly since 1990-1992, with gaps hovering at around 20-25 points for Hispanics and 25-30 points for Blacks.

Instead of comparing average scores between racial groups, it would be better to compare the percentage of students from each racial group that score at the highest levels because elite universities predominantly select from students performing at these levels. To get the percentage of students at the highest level, I will use the percentage of 12th-graders scoring at “advanced”, as these are the highest levels recognized by the NAEP. Scoring at these levels reflects the ability to perform the following tasks:

  • For mathematics [archived], this indicates the ability to use mathematical knowledge to “solve nonroutine and challenging problems, provide mathematical justifications for their solutions, and make generalizations and provide mathematical justifications for those generalizations”.
  • For reading [archived], this indicates an ability to “describe more abstract themes and ideas in the overall text” and “analyze both the meaning and the form of the text and explicitly support their analyses with specific examples from the text”.

The Data Explorer at the official NAEP website allows us to create custom data tables, which provides information on the percentage of students of each racial group scoring at various achievement levels. In 2019, these were the percentage of 12th grade students scoring at the proficient or advanced level in reading and mathematics:

  • Reading: About 9% of whites who scored at or above advanced. This was about 9 times the percentage of blacks (1%) and 3 times the percentage of Hispanics (3%) who achieved the same level. Multi-racial (10%) and Asian (13%) students were slightly more likely than whites to score at the advanced level.

  • Mathematics: About 4% of whites scored at or above advanced. This was about 4 times the percentage of Hispanics who achieved the same (1%). Asians were actually about 3 times as likely as whites to score advanced or higher (14%). Multi-racial students were about equally as likely as whites to achieve this performance level (4%). The percentage of blacks who scored at advanced was so low that it was rounded to zero, so all we know is that less than 0.5% of blacks achieved this level. Since the percentage of whites who scored at advanced is at least 3.5%, this implies that whites were at least 7 times as likely as blacks to score at this level (>3.5% vs <0.5%).

  • Note: some data on the percentage of high-scoring students is presented on the predefined data tables at the NCES website (see here for reading and here for mathematics). But this data does not contain scores for Asians specifically (they are lumped in with “other”) and the reading data does not contain the percentage of students scoring at the highest level (350 points or higher).

In summary, nationally representative data shows massive racial differences in the percentage of students scoring at the highest level of academic achievement. Whites were about equally as likely as multi-racial students to reach the advanced level in both mathematics and reading. Asians were slightly (44%) more likely to reach this level in reading, but were 3-4 times more likely to reach this level in mathematics. Whites were about 3-4 times as likely as Hispanics and 7-9 times as likely as blacks to reach the highest levels in mathematics and reading.

SAT scores

The SAT also shows large racial gaps in test scores. In 2018, the mean total SAT score for white high school graduates was 1123 points (page 3, CollegeBoard 2018 [archived]). The mean score for Asians was 1223, 100 points higher than the score for whites. The mean scores for blacks and Hispanics were 946 and 990 points, respectively, about 130-180 points lower than the white mean. Multi-racial graduates scored at 1101, just 22 points lower than whites.

Just like with the NAEP score data, it will be useful to compare the percentages of each racial group scoring at the highest levels, since this captures the kinds of students who attend elite universities. Data suggest that the middle 50% of test scores range from the low 700s to the high 700s on the Reading/Writing and math sections. For example, the 25th and 75th percentile for SAT math scores are 740 and 800, respectively, at Princeton University. The same percentile for the SAT Reading/Writing scores are 710 and 770. There are similar ranges at UPennBrownVanderbilt, and Rice. Thus, I will compare racial differences in the percentage of students scoring above a 700 on the mathematics and Reading/Writing sections of the SAT.

Unsurprisingly, there were large racial differences in students who scored within this high range (see page 5 of the SAT source):

  • On the Evidence-Based Reading/Writing (ERW) section, 9% of white graduates scored in the 700-800 point range, which is about the same as the percentage of multi-racial graduates who scored the same (10%). By contrast, only 1% and 2% of black and Hispanic graduates, respectively, scored in this elite range. 16% of Asian graduates scored in this range, about 78% more than white graduates.
  • On the Math section, 10% of both white and multi-racial graduates scored in the 700-800 point range. By contrast, only 1% and 3% of black and Hispanic graduates, respectively, scored in this range. 36% of Asian graduates achieved a score in this range, over 3 times as many white graduates.

These patterns mostly match the patterns seen with the NAEP scores. White and multi-racial graduates were about equally as likely to achieve the highest level. Asian graduates were more likely to reach the highest level, with the Asian advantage being particularly pronounced in mathematics. Whites were about 3-4 times as likely as Hispanics and about 9-10 times as likely as blacks to achieve this level.

The math gaps are so pronounced that if you focus on those at the highest levels, the sample of students becomes overwhelmingly Asian and almost devoid of blacks and (to a lesser degree) Hispanics. For example, Reeves and Halikias (2017) [archived] noted the following about race gaps at the tails of SAT math scores:

Race gaps on the SATs are especially pronounced at the tails of the distribution. In a perfectly equal distribution, the racial breakdown of scores at every point in the distribution would mirror the composition of test-takers as whole i.e. 51 percent white, 21 percent Latino, 14 percent black, and 14 percent Asian. But in fact, among top scorers—those scoring between a 750 and 800—60 percent are Asian and 33 percent are white, compared to 5 percent Latino and 2 percent black. Meanwhile, among those scoring between 300 and 350, 37 percent are Latino, 35 percent are black, 21 percent are white, and 6 percent are Asian.

These gaps have been present for as long as SAT scores have been recorded. For example, a 2005 article [archived] published in The Journal of Blacks in Higher Education reported that there were “almost no blacks” among the top scorers on the SAT:

If we eliminate Asians and other minorities from the statistics and compare just white and black students, we find that 5.8 percent of all white SAT test takers scored 700 or above on the verbal portion of the test. But only 0.79 percent of all black SAT test takers scored at this level. Therefore, whites were more than seven times as likely as blacks to score 700 or above on the verbal SAT. Overall, there are more than 39 times as many whites as blacks who scored at least 700 on the verbal SAT.

On the math SAT, only 0.7 percent of all black test takers scored at least 700 compared to 6.3 percent of all white test takers. Thus, whites were nine times as likely as blacks to score 700 or above on the math SAT. Overall, there were 45 times as many whites as blacks who scored 700 or above on the math SAT.

If we raise the top-scoring threshold to students scoring 750 or above on both the math and verbal SAT — a level equal to the mean score of students entering the nation’s most selective colleges such as Harvard, Princeton, and CalTech — we find that in the entire country 244 blacks scored 750 or above on the math SAT and 363 black students scored 750 or above on the verbal portion of the test. Nationwide, 33,841 students scored at least 750 on the math test and 30,479 scored at least 750 on the verbal SAT. Therefore, black students made up 0.7 percent of the test takers who scored 750 or above on the math test and 1.2 percent of all test takers who scored 750 or above on the verbal section.

In summary, regardless of whether we use NAEP or SAT data, we find the same basic patterns. Whites are about 3-4 times as likely as Hispanics and about 8-10 times as likely as blacks to score at the highest levels. They are roughly equally as likely as multi-racial students to score at this level. Asians were slightly more likely than whites to score at the highest level in reading assessments, but were about 3 times as likely to score at the highest level in mathematics assessments.

Jewish analysis

As stated before, racial/ethnic classifications should capture the different social realities of different groups of people. One important difference in social realities that is not captured by the above data is that between Jewish and non-Jewish whites. About 85% of Jews aged 18-29 in the United States identify as white (Pew Research 2021 [archived]). So it’s fair to infer that the vast majority of Jewish students are grouped as “white” in the above statistics on enrollment in elite universities. That being said, we ought to separate Jewish and non-Jewish whites in these statistics, for the same reason we ought to separate Hispanic and non-Hispanic whites, i.e. Jewish and non-Jewish whites have vastly different social realities (as I show in the next section). Thus, at the end of this section, I’ll provide estimates for racial demographics at elite universities after separating Jewish and non-Jewish whites.

Jewish vs non-Jewish whites

Pew Research has provided a lot of good data on the economic, social, and political outcomes for Jews. The data shows that Jews are vastly different from non-Jews in each of these respects. For example, there are sharp economic differences between Jewish and non-Jewish households (Pew Research 2016 [archived]). Families that were religiously Jewish had higher household incomes than every other religious group in America. 44% of Jewish households had incomes exceeding $100,000. Hindus had the second highest household income, with only 36% of households exceeding that mark. 19% of households in the general population exceeded $100,000 in annual income.

There are also many other points of differences between the Jewish and non-Jewish population:

  • Similar data was published in a more recent report by Pew Research (2021) [archived]. This data showed that 54% of Jews live in households earning at least $100,000. By contrast, only 18-21% of Catholic/Protestant households achieved the same. Moreover, 23% of Jewish households earned at least $200,000, compared to only 3-4% of Catholic/Protestant households.
  • There are also sharp differences between the Jewish and non-Jewish population in terms of feelings of connectedness (Pew Research 2021 [archived]). About half (48%) of U.S. Jews feel “a great deal” of belonging to the Jewish people. The vast majority (85%) feel at least “some” belonging. Furthermore, about 28% of U.S. Jews feel “a great deal” of responsibility to help Jews around the world. The vast majority (79%) feel at least “some” responsibility to do the same. Jews reported making at least one financial donation to a Jewish charity or cause in the past year. Presumably, all of these same numbers will be far lower for non-Jewish whites.
  • Jews differ sharply from non-Jewish whites in terms of perceptions and experiences of anti-Semitism in the U.S. today (Pew Research 2021 [archived]). For example, half of Jews say they have experienced at least one form of anti-Semitism in the past year. Jews are also much more likely to perceive discrimination in the country. For example, 43% of Jews say there is a lot of discrimination against Jews in our society today, whereas less than 18% of white protestants say the same.
  • Finally, Jews differ sharply from non-Jewish whites in their political attitudes as well (Pew Research 2021 [archived]). 71% of Jews identify with the Democratic party, whereas only 15-36% of white protestants do the same. 50% of Jews describe themselves as liberal, whereas only 4-16% of white protestants do the same. Other surveys also show a minority of whites identify as Democrat (26%) or liberal (23%).

Jews are also highly over-represented at extreme levels of intellectual achievement. For example, despite constituting just 0.2% of the world population (source), Jews have been awarded at least 20% of Nobel prizes (source). Thus, one should expect that Jews are also highly over-represented at America’s most elite colleges. Indeed, this is exactly the case. A 2018 piece in Tablet magazine that, while the proportion of Jewish students at Ivies universities has diminished in recent years, Jews still constitute somewhere between 10 to 20 percent of Ivies (despite constituting only 2% of the general population).

For example, data [archived] on religious demographics by Yale shows that about 16% of incoming freshmen reported Jewish affiliation during the 2010s:

At Harvard [archived], about 10% of incoming freshmen reported Jewish religious affiliation during the years 2017-2019, though that percentage dropped to 6% in 2020:

Estimating Jewish representation

Given the fact that Jews almost always identify as white, that they have vastly different social realities as whites, and that they are highly over-represented in elite universities, it is worth analyzing the racial demographics of elite universities after separating Jewish and non-Jewish (Gentile) whites. Perhaps the best resource for measuring Jewish representation comes from Hillel, an organization dedicated to fostering networks of Jewish college students. Hillel is active in hundreds of colleges and universities around the planet. Aside from just two schools (Notre Dame and Caltech), each of the top schools mentioned here have estimates of the Jewish population that have been provided by their respective local Hillel professionals. The estimates for their schools mostly line up with the numbers reported officially by universities. Consider the following examples:

  • Hillel estimates that about 10% of Harvard [archived] undergraduates are Jewish, which is in line with Harvard’s own estimate of the percentage of incoming freshmen that identifying as religiously Jewish (10% during 2017-2019, but only 6% in 2020).
  • Hillel estimates that 13% of Yale [archived] undergraduates are Jewish, in line with Yale’s own estimate of the percentage of incoming undergraduates proclaiming to be religiously Jewish. (16%).
  • Hillel estimates that 17% of undergraduates at the University of Pennsylvania [archived] are Jewish, in line with UPenn’s own estimate of the percentage of incoming 1st year students with Jewish religious affiliation (15%).

Estimates from Hillel suggest that about 8-18% of the undergraduates at these elite universities are Jewish. For each university, I estimate the number of Jewish and Gentile whites as follows.

  • Because about 85% of Jews aged 18-29 identify as white (Pew 2021 [archived]), I will assume that 85% of the Jewish undergraduates at each university are white. This provides an estimate of the number of white Jewish undergraduates at each university.
  • Once I have an estimate of the number of white Jewish undergraduates at each university, I estimate the number of Gentile whites by simply substracting this number from the total number of white undergraduates (reported above). For example, the above data showed that 38% of Harvard undergraduates are white. Hillel data estimates that 10% of Harvard undergraduates are Jewish. Assuming 85% of this 10% are white, this suggests that about 8.5% of Harvard undergraduates are white Jewish. Thus, about 8.5% of Harvard undergraduates are Jewish whites and 29.5% are Gentile whites.

Performing these calculations for each of the top universities reveals the following results:

Johns Hopkins7%21%26%16%8%7%0%3%12%
Notre Dame0%68%5%11%3%6%0%1%6%

The next table shows the summary statistics using the same calculations that I used for summary statistics that I presented earlier. Again, all of these calculations are published in a Google spreadsheet available here.

Weighted average (T20)9.9%26.7%21.8%14.7%6.5%6.1%
Weighted Average (Ivy only)12.5%24.8%20.2%12.6%7.7%5.6%
General Population (18 – 24 yo)2.04%50.9%5.7%23.1%14.0%3.3%
Enrollment Rate / White Enrollment Rate (T20)
Enrollment Rate / White Enrollment Rate (Ivy)12.621.

The 2.04% comes from the fact that 2.4% of the general population is Jewish and 85% of Jews aged 18-29 are white (Pew 2021 [archived]). So about 85% * 2.4% = 2.04% of the college-aged population is expected to be Jewish white. Now, this calculation makes two assumptions that I find to be reasonable:

  • The percentage of the college-aged population (i.e., ~18-24 yo) that is Jewish is roughly equal to the percentage of the general population that is Jewish (2.4%).
  • The percentage of Jews who are white among the college-aged population is roughly equal to the percentage of Jews who are white among elite undergraduates (85%).

These findings are much more surprising than the data presented above. Some important findings include:

  • Only 27% of T20 undergraduates and 25% of Ivy League undergraduates are Gentile whites. This is despite the fact that Gentile whites can be expected to make up about 51% of the college-aged population.
  • Jewish, Asian, and multi-racial students have enrollment rates at T20 universities that are about 9 times, 7 times, and 4 times greater than the same rate for Gentile white students. At Ivy Leagues in particular, Jewish whites have enrollment rates about 13 times as large as the same rate for Gentile whites. The data suggests that Gentile whites are actually severely under-represented.

In fact, Gentile whites seem to be even more under-represented than blacks and Hispanics at elite universities. This data suggests that, among the top 20 universities, the enrollment rate for Hispanics is actually 21% greater than the same rate for Gentile whites. The same rate for blacks is a mere 11% lower than the rate for Gentile whites. Among Ivy League universities, the enrollment rates for both black and Hispanic students are greater than the rate for Gentile whites by about 12-13%. This data suggests that Gentile whites are currently the most under-represented of the major racial/ethnic groups at Ivy League universities.

I must emphasize that these are the unadjusted enrollment rates. That is, these enrollment rates have not been adjusted for GPA, standardized test scores, or even parental socioeconomic status, which one would expect to cause much greater enrollment rates for whites than blacks/Hispanics. Racial preferences at these universities have not just reduced racial gaps in elite college enrollment; they’ve completely reversed the gaps.

Measuring racial preferences

The previous section hinted at racial preferences. This section will cite studies that attempt to precisely quantify the magnitude of racial preferences. Before citing this evidence, I would like to emphasize how unexpected the racial demographics of elite universities are given the racial demographics of the highest performers in high school.

First, note that the disparity between Asians and whites at elite universities is not surprising given differences in academic achievement cited earlier. Now, Asians are about 5.53 times as likely as whites to enroll at a T20 university, even though they were only 3 times as likely to score at the highest academic levels. But this difference isn’t too surprising because it may be the case that there are 5-6 times as many Asians at the extremely high levels of achievement (e.g., among those achieving perfect SAT scores), since the Asian predominance increases as the performance level increases.

On the other hand, the prevalence of Hispanics, blacks, and multi-racial students at elite universities is very surprising. Consider the following:

  • Multi-racial students are about equally as likely as whites to score at the highest levels, but they are 2.66 times as likely as white students to enroll at a T20 university.
  • White students are about 3-4 times as likely as Hispanic students to score at the highest levels, yet they are only about 8% more likely to enroll at a T20 university. There are about 6-13 times as many whites as Hispanics scoring above 700 on the math section of the SAT (see Brookings source above), yet there are only about 2.5 times as many whites as Hispanics enrolled at T20 universities.
  • White students are about 8-10 times as likely as black students to score at the highest levels, yet they are only about 33% more likely to enroll at a T20 university. There are about 16-27 times as many whites as blacks scoring above 700 on the math section of the SAT, yet there are only about 6 times as many whites as blacks enrolled at T20 universities.

Note that these figures include Jewish students in the “white” category. The slight white advantage in enrollment rates would be reversed if we specifically compared Gentile whites to the non-white racial groups.

Now, it is possible that multi-racial students, perhaps like Asians, are several times more likely to achieve at the extremely high levels than whites (e.g., perhaps there are a lot of Asian-white mixed children with perfect SAT scores), and perhaps this explains the racial disparity in elite university enrollment. That’s possible, but we would need to see more data to verify this. Regardless, there is only one explanation for the high enrollment rates for Hispanic and black students relative to whites (given the racial differences in academic achievement): racial preferences.

Most people know that selective universities display preferences for black and Hispanic applicants during the admissions process, but I don’t think people are aware of the magnitude of these preferences. One way to show the magnitude of racial preferences involves comparing the demographics of students at elite universities to the demographics of high school students with the highest level of academic achievement (as I did earlier). Another way involves simulating counterfactual demographics of selective universities assuming there were no racial preferences. I will cite studies doing just this in the rest of the section. I will also cite data comparing the magnitude of race-based preferences to the magnitude of class-based preferences.

Before citing that data, it is important to note that, as Arcidiacono et al. (2010) [archived] have reported, “only the most selective colleges employ racial admissions preferences” (page 2). In fact, racial preferences at selective universities have resulted in a U-shaped relationship between the average SAT score of an institution and the percentage of undergraduates that are black (Figure 1).

Thus, even though black students have far lower test scores than average, “the most selective schools have a higher percent black than moderately selective schools” (page 6). In other words, as you move from middling schools to the best schools, the percentage of students who are black increases, which is a remarkable finding given the abysmally low numbers of black high school graduates reaching the highest level of academic achievement.

That being said, let us review data that attempts to measure the magnitude of racial preferences at selective institutions.

Case Studies

Harvard University is a good case study to examine the effects of racial preferences since they have recently been involved in lawsuits regarding their implementation of affirmative action. In one lawsuit [archived], the Students for Fair Admissions (SFFA) presented data revealing “astonishing racial disparities in admission rates among similarly qualified applicants” (page 11). They measured academic qualification using an “academic index” created by Harvard’s method for indexing test scores and GPA. They found that, among applicants with the same academic index, Asian applicants had lower admit rates than white applicants, who had far lower admit rates than Hispanic and black applicants. For example, the case notes (page 11):

For example, an Asian American in the fourth-lowest decile has virtually no chance of being admitted to Harvard (0.9%); but an African American in that decile has a higher chance of admission (12.8%) than an Asian American in the top decile (12.7%).

They estimate that nearly half of Hispanics and black students wouldn’t be present if not for racial preferences (page 12):

SFFA’s regression analysis showed “substantial” preferences for African-American and Hispanic applicants. JA.2290:22-2291:8; JA.6017. Harvard’s expert, David Card, agreed. If Harvard eliminated racial preferences and adopted no race-neutral alternatives, Card found that the African-American share of the class would fall from 14% to 6% and the Hispanic share would fall from 14% to 9%. App.209-10; JA.6121. In absolute terms, then, race was “determinative” for at least “45% of all admitted African American and Hispanic applicants”—or “nearly 1,000 students” over a four-year period. App.209.

Of course, Harvard also admits many students based on legacy status. So one might wonder if blacks and Hispanics would have lower enrollment rates if we removed racial preferences and legacy preferences. Arcidiacono et al. (2020) [archived] performed an analysis to answer this question. Researchers found that eliminating preferences for legacies and athletes would significantly reduce the share of white admits and slightly increase the share of black and Hispanic admits. However, removing legacy/athlete preferences and racial preferences would significantly reduce the share of black/Hispanic admits, slightly increase the share of white admits, and significantly increase the share of Asian American admits. The researchers predict the following would occur under such a system (page 17):

The last row in Table 5 shows what would happen if in addition to removing legacy and athlete preferences, we also removed racial preferences. In this case, the coefficients on legacy, athlete, and race/ethnicity are set to zero as well as their interactions. The counterfactual shows that the number of admitted African Americans would be a third of what it was when all these preferences were in place. The number of admitted Hispanics would decline by almost half. Clearly the preferences African Americans and Hispanics receive do not simply offset the losses they incur from legacy and athlete preferences.

Their simulations estimate that the number of whites would increase slightly (from 4,802 to 4,947) and the number of Asians would increase significantly (from 2,358 to 3,564).

Arcidiacono et al. (2014) [archived] also estimated the magnitude of racial preferences at another highly selective university, the University of North Carolina at Chapel Hill. Unsurprisingly, black and Hispanic applicants had lower scores than Asian and white applicants (Table 1). Despite having lower test scores, the admission rates for black (40.9%) and Hispanic (52.5%) applicants were greater than the admission rates for white (38.2%) and Asian (33.5%) applicants.

The magnitude of racial preferences was estimated by running a regression analysis on whether an individual applicant was admitted with independent variables being SAT scores, an indicator for high school grade bins, racial/ethnic group, and sex. Consistent with the evidence from the Harvard case, large racial preferences were observed for black and Hispanic applicants. The magnitude of the racial preferences were equivalent to the effect of hundreds of SAT points (page 495):

To get a better sense of the degree of racial preferences, we estimate a logistic regression, in which the dependent variable is whether the individual was admitted and the covariates are SAT scores, indicators for different high school grade bins, indicators for different racial/ethnic groups, and an indicator for being male. Results are presented in Table 2. Converting the coefficients on race/ethnicity to SAT scores and assuming that any measures of unobserved preparation and character-related aspects of admissions are the same across groups, we find that preferences for African Americans (Hispanics) are equivalent to 352 (241) points on the SAT. Asian students, however, suffer a 46-point penalty. The logit coefficients also allow us to construct an academic index, in which we weight high school grades and SAT scores by treating the logit coefficients as weights that tell us the relative importance of each factor. The median African American (Hispanic) admit had an index value less than the 7.5th (17.5th) percentile of the white admit distribution. We note that preferences are likely understated here owing to selection on unobservables. Given knowledge of racial preferences, weaker minority applicants may apply knowing that they still have a chance of being admitted while nonminority students with the same credentials may not because of the lower probability of a positive outcome. Because our measures of academic preparation are imperfect, this effect biases our coefficients on race/ethnicity downward, understating the degree of racial preferences

The researchers ran simulations to predict what admission rates would be for each racial group if there were no racial preferences. Assuming the applicant pool was unchanged, the simulation indicated that the admission rate for black applicants would drop substantially, from 40.9% to 13.5%. The admission rates for Hispanic applicants would also drop substantially, from 52.5% to 29%. The data suggest that between about one-half and two-thirds of the black and Hispanic admits at this university would not have been admitted absent racial preferences.

Racial preferences vs class preferences

Given the large racial preferences exercised by elite universities, one might want to compare the magnitude of preferences for under-represented minorities to the magnitude of preferences for poor and low-SES applicants. In this section, I’ll present data analyzing larger samples of selective universities. This data attempts to quantify the magnitude of race-based affirmative action and compares this to the magnitude of any class-based affirmative action. The majority of the data finds large racial preferences, but little to no preferences for poor or low-SES applicants.

Consider an analysis of selective college admissions by Carnevale and Rose (2003) [archived]. These researchers used information from two sets of longitudinal data published by the National Center for Education Statistics (NCES) — the National Education Longitudinal Study of 1988 (NELS:88) and the High School and Beyond study (HS&B 80:92) — to analyze trends in admissions decisions between 1979 and 2000. Of particular interest here is the researchers’ analysis of the “top tier” competitive colleges. This includes 146 four-year colleges (less than 10% of the country’s four-year institutions), generally accepting less than 50% of their applicants. Researchers found significant racial preferences at these elite universities, but no preference for low-income students. The authors conclude that “on average the top 146 colleges do not provide a systemic preference, and could in fact admit far greater numbers of low-income students, including low-income minority students” (page 6).

The authors simulated possible admissions outcomes if decisions were based “exclusively on the most easily quantifiable academic measures, including grades, college entrance exam scores, teacher recommendation, and participation and leadership in extracurricular activities” (page 44). This simulation suggests that current admissions decisions by elite universities results in far more black and Hispanic students than what would be expected based solely on these “easily quantifiable academic measures”, but there are actually slightly less low-income students than what would be expected on that same scheme (page 37):

Given broad societal agreement among the public and college admissions officers that merit should be defined partly in terms of difficulties overcome, what sort of consideration of obstacles—racial and socioeconomic—is in fact given? Our own analysis finds that race and ethnicity is a significant consideration for colleges, boosting admissions from 4 percent under a system strictly of grades and test scores for African Americans and Latinos to 12 percent enrolled. By contrast, being economically disadvantaged, on net, reduces rather than improves chances of enrolling at one of the 146 most selective colleges. Admission based on tests and grades alone increases socioeconomic diversity marginally, from the current 9 to 12 percent from the bottom half of the income distribution.

In other words, if admissions decisions were based solely on academic measures, selective universities would admit slightly more low-income students and significantly less black and Hispanic students.

In chapter 3 of No Longer Separate, Not Yet Equal (2009), Princeton sociologists Thomas Espenshade and Alexandria Radford attempt to estimate advantages for minority applicants in seven elite American universities in the fall of 1997. The researchers used data from the National Survey of College Experience. The authors found that the advantage associated with black applicants is equivalent to 310 SAT points at private institutions and 3.8 ACT points at public institutions (page 93):

The second column of Table 3.5 indicates the size of admission preferences at private NSCE institutions. Once again, black applicants receive the largest admission bonus—equivalent to 310 SAT points. A black candidate with an SAT score of 1250 could be expected to have the same chance of being admitted as a white student whose SAT score is 1560, all other things equal in model 5. The average admission preference accorded to Hispanic applicants is roughly the same as having an extra 130 SAT points. On the other hand, an Asian candidate with a 1250 SAT score would be just as likely to be admitted at a private NSCE institution as a white student with an SAT score of 1110, other things the same.

  • Note: the authors of the book deny that this data implies that black applicants have an advantage because they are black. They state “it would be a mistake to interpret the data in Table 3.5 as meaning that elite college admission officers are necessarily giving extra weight to black and Hispanic candidates just because they belong to underrepresented minority groups” (page 94). In their judgment, the advantage for black and Hispanic applicants occurs because race is a proxy for a constellation of other unobserved factors that correlate with race, such as “perhaps having overcome disadvantage and limited opportunity or experiencing challenging family or schooling circumstances”. I’ll leave it to the reader to determine if this judgment seems plausible.

To illustrate just how large this advantage is, let us use the example that the authors provide, i.e. that a black candidate with an SAT score of 1250 has the same chance of admission as a white student with an SAT score of 1560. Using data on SAT percentiles here, these scores correspond to the 81st and 99th percentiles, respectively, of SAT takers. Thus, these scores imply that a black student scoring at the 81st percentile of SAT test-takers has about the same chances of admission as a white student scoring at the 99th percentile, an extremely large difference. A gap of 310 points also corresponds to the gap between a score at the 90th percentile of test-takers (1350 points) and a score at the 49th percentile of test takers (1040 points).

To put the black advantage into context, the advantage associated with being black was nearly 3 times greater than the advantage associated with being lower class at private institutions. At public institutions, there was no advantage for lower class applicants. Furthermore, the researchers found that the slight advantage for lower class applicants at private institutions was limited to non-white applicants. In fact, lower class white applicants were at a disadvantage, relative to higher class white applicants (page 100):

This conclusion of no low SES admission preference is corroborated quite well by our findings at public NSCE universities (see Figure 3.8). Within each racial or ethnic group of applicants, the predicted probability of being admitted, on an all-other-things-equal basis, varies relatively little from one social class to another. It is clear that applicants to public universities who come from lower- and working-class family backgrounds do not have an admission advantage compared with other applicants. But we cannot conclude from our findings at private NSCE colleges and universities that there is no SES effect. Figure 3.9 shows that for white applicants to private institutions, there is a low SES admissions disadvantage. White applicants from lower- and working-class families have admissions chances of 8 percent and 18 percent, respectively, in contrast to an expected chance of being admitted that falls in the 20–30 percent range for students from higher SES families. For nonwhite students, on the other hand, there are clear signs of a low SES admissions advantage. Black applicants who come from lower- or working-class families can expect a favorable admissions decision in 87 percent and 53 percent of their cases, respectively. The expected chance of being admitted falls to just 17 percent for upper-class black students. Strikingly similar patterns characterize the admissions chances by social class for Hispanic and Asian applicants to private institutions.

Surprisingly, at private institutions, lower class whites had the lowest chances of admissions out of all race-class groups (after controlling for other factors; for a full list of all controls, view Table 3.5 posted above). Lower class blacks were about 10 times as likely as lower class whites to be admitted to private institutions. Even upper middle class blacks were over 5 times as likely as lower class whites to be admitted.

Reardon et al. (2017) [archived] produced simulations to estimate the extent to which SES-based affirmative action could produce racial diversity in selective colleges. Researchers ran simulations based on data in the Education Longitudinal Survey of 2002, (ELS), a “nationally representative sample of high school students who would graduate in 2004” (page 14). Consistent with prior research, the authors find that highly selective universities have large preferences for black and Hispanic applicants with very small preferences for low-SES students. The findings from the study are summarized as follows (page 12):

The results of our analyses suggest that Black and Hispanic applicants to the most selective colleges receive an implicit admissions weight that is roughly equal to the weight given to a 1.3 standard deviation increase in academic performance (in other words, the difference in the probability of admission of White and Black or Hispanic students is roughly equal to the difference in the probability of admission of two students of the same race whose academic performance differs by 1.3 standard deviations). We find very little or no evidence of racial preferences in admissions to colleges in lower selectivity tiers (see Appendix Table B1).

We find evidence of slight SES-based affirmative action in the most selective colleges—a standard deviation difference in family SES is roughly the same as a 0.15 standard deviation difference in academic record. However, lower-SES students applying to less selective colleges appear to be penalized in the admission process. In these colleges higher SES students were given implicit preference in admissions decisions. The SES weights are, however, relatively small in all cases. This heterogeneity perhaps reflects the fact that existing SES-based admissions preferences work in two directions: On the one hand, most colleges rely heavily on student tuition and must take ability to pay into account in admissions; on the other hand, many colleges, particularly very selective colleges, actively recruit and admit low-SES students (see Appendix Table B2).

The study concluded with the following (page 54):

In sum, it appears that, in 2004, affirmative action or other related policies at the most selective
colleges increased the probability of minority students’ admission substantially by an amount that may be as high as the difference between students whose academic records differ by over a standard deviation. SES-based affirmative action policies, however, appear to have been much less prevalent. On average, low-SES applicants appear to have received little or no admissions preference at most colleges.

In a recent review of admissions practices at selective colleges, Giancola and Kahlenberg (2016) [archived] cite some research on the magnitude of the non-academic admissions preferences. These preferences are based on many factors, including athletic ability, being an under-represented minority, having a relative who attended the institution (“legacy” applicant), and having sufficient means to pay full tuition. Students with these preferential factors are described as being “hooked” (page 23). While each of these students are given significant preferences, there are no preferences given to low-income students (page 24):

Race-conscious affirmative action has been used for decades to address past inequities and offer students from disadvantaged minority groups (particularly African Americans and Latinos) a better chance at gaining access to college. As a result, in selective colleges and universities today, underrepresented minorities receive a 28 percentage point boost in admissions compared to White students with the same credentials. Low-income students who are not minorities receive no such advantage

In fact, one university chancellor estimated that only 40% of the available seats at Ivy League institutions are “unhooked” (page 27). Researchers note that the effect of all these non-academic preferences leaves little room for high-achieving, low-income students who do not possess the preferred qualities (page 26):

By the time all of the admissions preferences and processes (such as early decision and demonstrated interest) are added up, there is little room left for high-achieving, low-income students, nor do these students receive any preferential treatment of their own for having overcome the often steep barriers that result from growing up in poverty. In fact, analysis of 13 selective institutions’ admissions data suggests that while being an athlete, under-represented minority, or legacy provides considerable advantage in college and university admissions, emerging from low-income families and being a top student provides no boost whatsoever (Figure 17).

The basic findings from these studies are consistent: selective universities engage in large preferences for black and Hispanic applicants, and little to no preference for poor applicants. For other articles reviewing mostly the same data on the relative effects of race-based affirmative action versus SES-based affirmative action, see Giancola and Kahlenberg (2016) [archived] and Kahlenberg (2015) [archived].

Immigrant preferences

We also find large preferences for immigrants who happen to be under-represented minorities. Black immigrants in particular are important because they constitute a substantial portion of black students at elite universities. For example, consider the findings of Massey et al. (2007) [archived] who used data from the National Longitudinal Survey of Freshmen (NLSF) to study black immigrants and natives attending selective colleges. The sample and findings of the study were as follows:

  • The NLSF is a nationwide survey of freshmen who entered 28 selective colleges and universities in the fall of 1999. The baseline sample had 1,051 blacks, 959 whites, 998 Asians, and 916 Hispanics.
  • This survey showed that about 36% of the black freshmen at the 10 most selective universities were of immigrant origin (Table 1). About 41% of black freshmen at the Ivy League institutions were of immigrant origin. Black students with at least one foreign-born parent were reported as being of immigrant origin. Thus, many black students not counted as immigrant origin could have had foreign-born grandparents.

Other data show that selective universities also have large preferences for black immigrants. Bennett and Lutz (2009) [archived] used data from the National Education Longitudinal Study of 1988 (NELS:88) to analyze whether the black advantage in college admissions extends to black immigrants. The sample and findings of the study were as follows:

  • The NELS:88 is a nationally representative sample of schools and 8th graders in 1988. The authors analyzed data from the base year in 1988 through the third follow-up in 1994, two years after high school graduation. They focused their analysis on 8,552 students, including 7,499 whites, 958 native (third- and later-generation) blacks, and 95 immigrant (first- and second-generation) blacks.
  • Prior to introducing any controls, immigrant blacks were more likely to attend selective colleges than both native blacks and whites. 9.2% of immigrant blacks attended selective colleges, compared to only 7.3% of whites and 2.4% of native blacks (Figure 1).
  • After controlling for a robust set of covariates, both native and immigrant blacks were far more likely to enroll in a selective university. In particular, the odds for enrollment at a selective university for native blacks and immigrant blacks was 3.7 times and 17 times, respectively, greater than the odds for whites (Panel D of Table 3 and Appendix D). The authors note the following (page 84):

Both black immigrants and native blacks evidence a net advantage over whites in enrollment in selective colleges (see Panel D). Whereas native blacks are 3.7 times more likely than are similar whites to attend selective colleges, black immigrants are many times (OR = 17.0) more likely than are whites to do so compared to not attending college at all.

The control variables included gender, family socioeconomic background (a composite reflecting family income, parental education, and parental occupation), family structure, region, private vs public school, standardized test scores, and expectations for a college degree (Model A from Table 2). Here is the full table of the results of the analysis described above:

Returning to Chapter 3 of No Longer Separate, Not Yet Equal (2009), researchers continued their analysis of the seven elite NSCE institutions by examining the influence of immigration status. This is important because a substantial portion of applicants came from 1st, 2nd or 3rd generation immigrants (31% for public institutions and 59% for private institutions, Table 3.6). In fact, among private institutions, about 40% of black and 66% of black and Hispanic applicants, respectively, were from 1st or 2nd generation immigrants. The authors also tried to separate blacks who are likely to descend from the slave population vs blacks who did not. They categorized and labelled these groups of blacks as follows:

  • “Descendants” is a phrase used by Harvard black students to refer to black students who descended from the American slave population (page 105). In this study, “descendants” refers to black people who are not multi-racial who are 4th or higher immigrant generation (this includes black natives). The “descendants” make up 68% of black applicants at public institutions and only 36% of black applicants at private institutions (page 105).

After disaggregating these racial groups by immigrant generation and “descendant” status, the admissions advantages for these groups were are as follows (Table 3.7):

  • Both black “descendants” and “non-descendants” benefited from large admissions preferences at both public and private institutions. The authors note that the significantly positive preference for black applicants observed “is traceable to additional weight given to both black descendants and black non-descendants compared to white students” (page 107).
  • Recall that Hispanic applicants received admissions preferences in this study only in private institutions. The results indicated positive preferences for all generations of Hispanic immigrants, but only the advantages for second generation Hispanics was statistically significant. The authors note that “admission advantage for Hispanic applicants at private institutions is associated most closely with candidates in the second generation. The odds of being admitted for these students are almost three times as high as they are for comparable white applicants” (page 108).
  • After disaggregating racial groups by immigrant generation and “descendant” status, there was no advantage associated with being lower class. The authors comment that “neither the high nor low social class category differs significantly from middle-class applicants in terms of the chances of being admitted” (page 107).

The following chart displays the advantages by race, immigration/descendant status, and social class at public institutions. The data shows that black descendants and non-descendants received similarly large advantages. In fact, both high-SES black descendants and non-descendants had larger chances of being admitted than low-SES white applicants, after holding constant all other factors.

The following chart displays the same information at private institutions. The data shows advantages for both black “descendants” and “non-descendants”, although the descendant advantage is noticeably greater in private institutions, particularly among higher-SES applicants. In fact, both black high-SES descendants and non-descendants had higher probabilities of being admitted than low-SES white applicants, holding all other factors constant. Hispanic applicants of all immigration statuses had advantages (relative to white applicants) among low-SES applicants. Among middle-SES applicants, second-generation Hispanic immigrants also had advantages relative to white applicants.

Explaining racial preferences

Given the large racial preferences of elite American universities, a natural question to ask is, why do elite universities impose such strong preferences for black and Hispanic applicants? This question is important because the answer reveals something important about the political vision of our elite institutions of higher education. And the political vision of our elite institutions of higher education is important because the graduates of these institutions will have tremendous influence on the political, economic, and cultural future of our country (as I show below).

Possible explanations

Now, if you hadn’t already read the first half of this post, one might explain the admissions advantage for black and Hispanic applicants by positing that this advantage is just a byproduct of the fact that (a) universities have strong preferences for low-SES applicants and (b) black and Hispanic applicants are much more likely to be low-SES. However, the data in the previous section shows that this theory is completely false. The basic findings from studies comparing the relative effects of race-based affirmative action and class-based affirmative action are clear: there are large preferences for under-represented minorities even after controlling for SES preferences, and these racial preferences (particularly for black applicants) are far stronger than preferences for low-SES applicants. In fact, many of the reviews cited above find no preference at all for low-SES applicants.

Another possible explanation for the admissions advantages of black and Hispanics is that they are intended to rectify for past discrimination. As I will show below, this may have been an adequate explanation of past uses of affirmative action. But there are a number of problems with using this theory to explain modern uses of affirmative action. Here are just a few problems with this explanation:

  • This doesn’t explain the large preferences given to recent black immigrants (see above). Not only are they not the intended beneficiaries of historical racism within the United States, but this group is already over-represented at elite universities (based on data by Bennett and Lutz 2009 showing that black immigrants are more likely to attend universities than native whites).
  • This doesn’t explain the large preferences given to Hispanics (see above), most of whom are the descendants of recent immigrants (i.e. post 1960s). In 1970, well after laws enforcing slavery, segregation, etc., there were only 9.6 million Hispanics in the United States. That number had risen to 62 million by 2020 (Pew Research 2021). The vast majority of this growth was the result of recent immigration. For example, Durand et al. (2006) reported that about half of the growth in the Hispanic population from 1970 to 2000 was due to foreign-born immigrants, and about a quarter was due to native-born children of foreign born immigrants (Table 3-1). In fact, in 2015, the vast majority (60%) of Hispanic children in the United States were either immigrants or the children of immigrants (Pew Research 2017).

But the biggest problem with this explanation of racial preferences is that it doesn’t line up with the justifications provided by universities in high-profile court cases. As I will show in the next section, the Supreme Court has increasingly rejected the notion that there’s a compelling government interest in adopting racial preferences to correct for societal discrimination. Instead, the justifications used by universities to defend affirmative action have relied on appeal to the educational benefits of diversity.

In order to explain racial preferences by elite universities, we should examine the actual justifications for racial preferences that are provided and accepted/rejected in high-profile court cases. These cases are important because the outcome of these cases set the standards for lawful use of racial preferences that other universities are obliged to follow. These cases generally involve public universities, but they also apply to selective private institutions since they receive federal funding. As Back and Hsin (2019) note, “Title VI of the Civil Rights Act of 1964 prohibits recipients of federal funding — including private colleges and universities — from, at a minimum, discriminating against students and applicants in a manner that would violate the Equal Protection Clause.” As it happens, elite private universities receive enormous funding by the federal government. In fact, a 2017 oversight report [archived] showed that “The eight colleges of the Ivy League received more money ($4.31 billion) – on average – annually from the federal government than sixteen states” (page 3).

Several reviews of affirmative action case law agree that the primary defenses of racial preferences shifted from appeal to a rectification principle (racial preferences are justified as a means to correct for historical or present discrimination) to appeal to a diversity principle (racial preferences are justified as a means to seek the educational benefits of diversity). Since I’m not expert on case law, I’ll just quote the relevant passages from various reviews by more informed writers in the field. For example, Massey et al. (2007) briefly described the changing rationales for racial preferences as follows (page 243):

Prior to the civil rights era, Americans of African origin were largely excluded from selective colleges and universities in the United States through a combination of de facto and de jure mechanisms. Once discrimination in education was definitively banned by the Civil Rights Act of 1964, however, things began to change. During the late 1960s, elite schools throughout the country began to undertake various “affirmative actions” to increase black enrollment. As outlined in the celebrated speech made by President Lyndon B. Johnson at Howard University, the initial justification for this policy was restitution for past wrongs:

You do not wipe away the scars of centuries by saying: Now you are free to go where you want, and do as you desire, and choose the leaders you please. You do not take a person who, for years, has been hobbled by chains and liberate him, bring him up to the starting line of a race and then say, “you are free to compete with all the others,” and still justly believe that you have been completely fair. Thus it is not enough just to open the gates of opportunity. All our citizens must have the ability to walk through those gates. This is the next and the more profound stage of the battle for civil rights. We seek not just freedom but opportunity. We seek not just legal equity but human ability, not just equality as a right and a theory but equality as a fact and equality as a result. ( Johnson 1965)

According to this rationale, the deliberate recruitment of African Americans into America’s top colleges and universities was justified to make up for generations of past exclusion. Soon, however, Latinos, Asians, women, and the disabled took note of the success of the civil rights movement and appropriated the tactics and rhetoric of African Americans to make their own demands for inclusion (Skrentny 2002). This broadening of the scope of civil rights coincided with a remarkable upsurge in immigration from Asia and Latin America, and over time the moral justification for affirmative action shifted subtly from restitution for a legacy of racism to the representation of diversity for its own sake (Graham 2002).

Another review of the history of affirmative action by Lehmuller and Gregory (2004) [archived] also noted the dwindling support by the Court of racial preferences to remedy past injustices, though the Court has maintained that there is compelling governmental interest to promote diversity (page 443):

In Wygant v. Jackson Board of Education (1986), the Court held that there must be a connection between past discrimination and the remedy (Hendrickson, 2001). As time passed, the appointment of more conservative Justices meant that a plurality of the Court no longer believed that government redress for past racial injustice was a compelling reason for sustaining affirmative action programs (Lewis, Lewis, & Ponterotto, 1990). In City of Richmond v. J.A. Crosson Co. (1989), the Court held that simply asserting an interest in remedying societal discrimination did not survive strict scrutiny (Grossett, 2001). In essence, any legislative attempt to correct past discrimination based on race must withstand strict scrutiny (Lewis et al., 1990 [emphasis added]). While affirmative action programs are permissible under the Fourteenth Amendment, any program must be “narrowly tailored to remedy identified discrimination” (p. 35).

Just 1 year later, however, the Court heard Metro Broadcasting v. FCC (1990), and held that the government did have a substantial interest in promoting diversity in the issuance of broadcast licenses (Grossett, 2001). The Metro decision was overruled in Adarand v. Peña (1995), although the Court did not disavow the concept of diversity as a compelling interest (Grossett).

The authors go on to note rising use of diversity as the primary rationale for racial preferences as other rationales were rejected by the courts (page 446):

Diversity as a compelling interest became increasingly important in affirmative action cases as other rationales failed to survive judicial review. Racial balancing through quotas was impermissible after Bakke, the “positive role model” theory was discredited in Wygant, legislatures could not make broad categories for remedy after Crosson, and affirmative action could not be considered permanent or a justification for proportional representation after Lutheran Church Missouri Synod v. FCC (1998) (Pavela, 2002).

Finally, another review of affirmative action case law by Sedler (2003) [archived] describes the specific propositions emerging from the Powell opinion Bakke v. Regents of the University of California, a landmark Supreme Court case in which the University of California was sued for their use of racial quotas. This opinion explicitly rejected the notion of a compelling governmental interest in using racial preferences to overcome general societal discrimination (though the opinion did note that there is compelling interest in a governmental agency overcoming discrimination that the specific agency is responsible for) (page 221):

In Bakke, neither the university nor, to the best of my recollection, any of the numerous proponents of affirmative action in their amicus curiae briefs, tried to justify affirmative action on diversity grounds. Rather, the thrust of the arguments in support of the university’s affirmative action program was that the state’s interest in overcoming the present consequences of past societal discrimination against racial minorities was compelling. The university and amici further argued that a reasonable racial quota, such as allocating to minorities sixteen of the one hundred seats in the entering class of the medical school of the University of California at Davis, was narrowly tailored to advance that interest. This was the position taken by Justices Brennan, White, Marshall, and Blackmun in Bakke, and it led them to conclude that the Davis Medical School’s use of race in determining admissions was constitutional. Four other Justices, in an opinion by Justice Stevens, took the position that Title VI prohibited any use of race in admissions by universities receiving federal funds, and so these Justices held that the medical school’s admissions program violated Title VI. This left it up to Justice Powell to cast the deciding vote.

It is clear in retrospect that the Powell opinion for the Court in Bakke resolved a number of issues relating to the constitutional permissibility of race-conscious affirmative action and represents the holdings of the Court on these issues today. Powell’s resolution of these issues constituted holdings in the case itself when his views were supported by four other Justices either in the “Brennan four” or “Stevens four” opinions; also, some of the dicta in Powell’s opinion became law when they were affirmed by a Court majority in subsequent cases. In any event, the following propositions emerged from the Powell opinion in Bakke and were established constitutional doctrine at the time of Grutter-Gratz.

One: The test under Title VI for the permissible use of race in university admissions is the same as the constitutional test, so whatever is permitted or prohibited for a public university as a constitutional matter is also permitted or prohibited under Title VI for a private university receiving federal funds.

Two: Strict scrutiny applies to racial classifications benefitting racial minorities in the same manner as it applies to racial classifications disadvantaging racial minorities.

Three: Overcoming the present effects of societal discrimination against racial minorities is not a compelling governmental interest justifying governmental action benefitting racial minorities and disadvantaging whites.

Four: Overcoming the present effects of identified unlawful racial discrimination for which the government agency itself is responsible is a compelling governmental interest justifying the use of narrowly tailored race-conscious criteria.

The impact of the Bakke case is later described as follows (page 225):

As a practical matter, Bakke settled the question of the constitutional permissibility of race-conscious university admissions. The Constitution permitted universities to adopt admissions programs that included the “competitive consideration of race and ethnic origin” as one of the factors in determining admission. After Bakke, universities revised their race-conscious admissions programs to comply with the approach set forth in the Powell opinion and justified their consideration of race in determining admissions as advancing a university’s compelling interest in attaining educational diversity. While the pre-Bakke justification — the strong public interest in increasing the number of minority lawyers and doctors — remained a major motivation for law school and medical school race-conscious admissions programs, the constitutionality of these programs could now be sustained under the different educational diversity justification. Similarly, following Bakke, most universities adopted programs taking race into account in all undergraduate, graduate school, and professional school admissions, and today racial diversity has become a hallmark of the university scene. This is all due to Justice Powell’s going beyond the arguments of the parties in Bakke and invoking the educational diversity justification for race-conscious admissions programs.

Important questions about racial preferences

So the massive racial preferences at elite universities are allegedly promoted for the sake of the educational benefits of diversity. I don’t have a strong stance on these benefits. But regardless of one’s views on the educational value of diversity, there are a number of important questions that seem to stand out out given the data presented earlier. I believe the following questions should stand out to anyone truly interested in promoting diversity for the sake of educational benefits:

  • How many applicants of under-represented minorities need to be admitted in order to acquire sufficient benefits of diversity? How is this number determined? How are the benefits of racial diversity weighed against the costs of admitting applicants with lower academic qualifications?
  • If elite universities have large preferences for under-represented minorities because of the educational benefits of diversity, why don’t they have similarly large preferences for applicants from under-represented socioeconomic classes or applicants with under-represented viewpoints? Why are there no preferences for socioeconomic or viewpoint diversity?
  • What is the timeline for racial preferences at elite universities? At what point will racial preferences no longer be needed? What are the conditions that must be met to indicate that racial preferences are no longer needed?

I don’t have good answers to these questions (and there may be answers out there that I haven’t seen). But I’ll explore all of them in turn in this section.

What is a “critical mass”?

Regarding the first set of questions, this was somewhat addressed in Grutter v. Bollinger (2003), a landmark court case against the University of Michigan Law School for their use of racial preferences. The court ultimately ruled that preferential treatment for under-represented minority groups is not unconstitutional so long as other factors are considered in the admissions process. During the case, Dennis Shields, the director of admissions, testified that he directed staff to consider race in admissions in order to ensure that a “critical mass critical mass of underrepresented minority students would be reached so as to realize the educational benefits of a diverse student body”. But what is “critical mass” and how is this number determined? Erica Munzel, who succeeded Shields as Director of Admissions, testified that “critical mass” means “meaningful numbers” or “meaningful representation”, which refers to “a number that encourages underrepresented minority students to participate in the classroom and not feel isolated”:

Erica Munzel, who succeeded Shields as Director of Admissions, testified that “ ‘critical mass’ ” means “ ‘meaningful numbers’ ” or “ ‘meaningful representation,’ ” which she understood to mean a number that encourages underrepresented minority students to participate in the classroom and not feel isolated. Id., at 208a—209a. Munzel stated there is no number, percentage, or range of numbers or percentages that constitute critical mass. Id., at 209a. Munzel also asserted that she must consider the race of applicants because a critical mass of underrepresented minority students could not be enrolled if admissions decisions were based primarily on undergraduate GPAs and LSAT scores. Ibid.

Jeffrey Lehman, Dean of the Law School, also testified that “critical mass means numbers such that underrepresented minority students do not feel isolated or like spokespersons for their race”:

The current Dean of the Law School, Jeffrey Lehman, also testified. Like the other Law School witnesses, Lehman did not quantify critical mass in terms of numbers or percentages. Id., at 211a. He indicated that critical mass means numbers such that underrepresented minority students do not feel isolated or like spokespersons for their race. Ibid. When asked about the extent to which race is considered in admissions, Lehman testified that it varies from one applicant to another. Ibid. In some cases, according to Lehman’s testimony, an applicant’s race may play no role, while in others it may be a “ ‘determinative’ ” factor. Ibid.

These responses are not very clear answers to the question. What is the number of minority students necessary for them to not feel isolated or like spokespersons? How was this number determined? This is unclear. Furthermore, the appeal to ensuring a “critical mass” seems inconsistent with data on the enrollment of other minorities, as Judge Rehnquist explained in his dissent (page 4):

From 1995 through 2000, the Law School admitted between 1,130 and 1,310 students. Of those, between 13 and 19 were Native American, between 91 and 108 were African-Americans, and between 47 and 56 were Hispanic. If the Law School is admitting between 91 and 108 African-Americans in order to achieve “critical mass”, thereby preventing African-American students from feeling “isolated or like spokespersons for their race”, one would think that a number of the same order of magnitude would be necessary to accomplish the same purpose for Hispanics and Native Americans. Similarly, even if all of the Native American applicants admitted in a given year matriculate, which the record demonstrates is not at all the case,* how can this possibly constitute a “critical mass” of Native Americans in a class of over 350 students? In order for this pattern of admission to be consistent with the Law School’s explanation of “critical mass”, one would have to believe that the objectives of “critical mass” offered by respondents are achieved with only half the number of Hispanics and one-sixth the number of Native Americans as compared to African-Americans. But respondents offer no race-specific reasons for such disparities. Instead, they simply emphasize the importance of achieving “critical mass”, without any explanation of why that concept is applied differently among the three underrepresented minority groups.

So it’s not even clear what a “critical mass” is, how it’s determined, nor why it seems to vary from racial group to racial group. We may never have adequate answers to these questions.

Why no socioeconomic diversity?

The question about socioeconomic diversity diversity is particularly important because there is empirical evidence that elite universities could achieve significantly more socioeconomic diversity with only slight reductions in racial diversity by adopting preferences for low-SES students. Consider the following examples:

  • In an analysis of 146 “top tier” four-year colleges, Carnevale and Rose (2003) ran simulations which estimated that, if universities replaced racial preferences with economic preferences in their admissions policies, this would “somewhat reduce racial and ethnic diversity and greatly expand socioeconomic diversity” (page 7). In particular, this hypothetical admissions scheme would increase the percentage of students in the bottom two SES quartiles from 10 percent to 38 percent (page 55), and the pool of black and Hispanic students would decrease from 12 percent to 10 percent (page 55).
  • In an analysis of different hypothetical admissions policies at the top 193 colleges, Kahlenberg (2015) estimated that, “under a program of class-based affirmative action” as an alternative to race-based affirmative action, socioeconomic diversity would rise, mean SAT scores would rise, and even the combined black and Hispanic representation would rise from 11 percent to 13 percent. This is also assuming that legacy preferences and athletic preferences are abandoned.

Thus, different simulations of admissions policies shows that top universities could significantly improve socioeconomic diversity with little to no cost to racial diversity. In fact, some simulations suggest that class-based affirmative action could even promote racial diversity as well, assuming other admissions preferences (legacy and athletic preferences) are abolished. So why do universities continue with current admissions policies?

The answer to this question may differ by institution, but analyzing Harvard’s answer to this question may be useful. In the Harvard lawsuit [archived] by the Students for Fair Admissions (SFFA), the SFFA presented simulation showing that if Harvard eliminated racial preferences and provided stronger advantages for socioeconomic disadvantaged applicants, this would greatly increase socioeconomic diversity, maintain the current proportion of under-represented minorities, and maintain the current academic qualifications of its student body (page 18). However, the presented simulation also involved eliminating preferences for “the children of donors, alumni, and faculty/staff” (page 17). In response, Harvard’s witnesses “insisted that it was “very important” to continue giving preferences to the children of donors, alumni, and staff/faculty” (page 18).

What’s the timeline for racial preferences?

Another important question regarding racial preferences concerns time frame. Affirmative action was never meant to be permanent. It was meant to be an unfortunate measure to temporarily promote racial diversity for the educational benefits. In fact, in Grutter v. Bollinger (2003) [archived], Justice O’Connor delivered the opinion of the Court as follows:

We are mindful, however, that “[a] core purpose of the Fourteenth Amendment was to do away with all governmentally imposed discrimination based on race.” Palmore v. Sidoti, 466 U.S. 429, 432 (1984). Accordingly, race-conscious admissions policies must be limited in time. This requirement reflects that racial classifications, however compelling their goals, are potentially so dangerous that they may be employed no more broadly than the interest demands. Enshrining a permanent justification for racial preferences would offend this fundamental equal protection principle. We see no reason to exempt race-conscious admissions programs from the requirement that all governmental use of race must have a logical end point. The Law School, too, concedes that all “race-conscious programs must have reasonable durational limits.” Brief for Respondents Bollinger et al. 32.

The opinion continued with the following:

The requirement that all race-conscious admissions programs have a termination point “assure[s] all citizens that the deviation from the norm of equal treatment of all racial and ethnic groups is a temporary matter, a measure taken in the service of the goal of equality itself.” Richmond v. J. A. Croson Co., 488 U.S., at 510 (plurality opinion); see also Nathanson & Bartnik, The Constitutionality of Preferential Treatment for Minority Applicants to Professional Schools, 58 Chicago Bar Rec. 282, 293 (May—June 1977) (“It would be a sad day indeed, were America to become a quota-ridden society, with each identifiable minority assigned proportional representation in every desirable walk of life. But that is not the rationale for programs of preferential treatment; the acid test of their justification will be their efficacy in eliminating the need for any racial or ethnic preferences at all”).

We take the Law School at its word that it would “like nothing better than to find a race-neutral admissions formula” and will terminate its race-conscious admissions program as soon as practicable. See Brief for Respondents Bollinger et al. 34; Bakke, supra, at 317—318 (opinion of Powell, J.) (presuming good faith of university officials in the absence of a showing to the contrary). It has been 25 years since Justice Powell first approved the use of race to further an interest in student body diversity in the context of public higher education. Since that time, the number of minority applicants with high grades and test scores has indeed increased. See Tr. of Oral Arg. 43. We expect that 25 years from now, the use of racial preferences will no longer be necessary to further the interest approved today.

This court ruling was 18 years ago today, so we’re only 7 years away from the point that the court predicted racial preferences to no longer be necessary. Unfortunately, racial preferences at elite universities show no signs of stopping or even slowing down. So when do elite universities expect that racial preferences will be eliminated? What conditions must be satisfied in order for racial preferences to be eliminated? We currently have no answer to these questions. And if the trend of increasing emphasis on “diversity” at elite institutions (academic and non-academic) continues, it looks like racial preferences will never be eliminated (if universities remain free to make their own choices).

Why these racial preferences matter

Racial preferences at elite universities are important because the graduates of elite universities will have pivotal influence on the political, economic, and cultural future of the country. All of our lives are heavily influenced by the political, economic, and cultural patterns of the country. Thus, the admissions policies of elite universities will have significant downstream effects on all of our lives, since these policies dictate future leaders of the nation.

A good review of the influence of the graduates of elite universities was presented by Wai et al. (2019) [archived]. “Elite” schools in the U.S. were universities with average standardized test scores in the top 1% of scorers relative to the general population (page 81). This resulted in 29 total schools. Researchers found that a substantial portion of influential persons attended one of these elite institutions. For example, graduates from these universities constituted 20% of House members, 41% of senators, 41% of federal judges, 41% of Forbes 500 CEOs, 43% of Forbes self-made billionaires, 44% of New York Times Editors/Writers, 50% of Wall Street Journal Editors/Writers, 56% of Forbes Powerful women, and 85% of Forbes Powerful men. Here’s the full list (see first column).

Unsurprisingly, all of the Supreme Court justices attended a law school from an elite school (as of November 19th, 2021). Specifically, 4 justices attended Harvard Law School (Breyer, Gorsuch, Kagan, and Roberts), 4 attended Yale Law School (Alito, Kavanaugh, Sotomayer, and Thomas), and 1 attended Notre Dame (Barrett).

In Who’s Running America? (2014), Thomas R. Dye documented the educational institutions attended by a narrower definition of “elite”. The definition of “elite” encompassed just 4,000 individuals (roughly) with enormous power described as follows (page 10):

These top positions, taken collectively, control over half of the nation’s industrial assets, over one-half of all US banking assets, and over three-quarters of all insurance assets. They control the television networks, the investment firms, the influential newspapers, and the major media conglomerates. They control over half of all the assets of private foundations and two-thirds of all private university endowments. They direct the nation’s largest and best- known law firms in New York and Washington, as well as the nation’s major civic and cultural organizations. They make the largest political campaign contributions. They occupy key federal government positions in the executive, legislative, and judicial branches (see Table 1.1).

Some of the components of the elite are described as follows (page 9):

  • In mass media, we include ownership of the major television networks (ABC, CBS, NBC, FOX, and CNN), the New York Times and Washington Post, and the nation’s five leading media conglomerates.
  • In law, our definition of the nation’s legal elite includes the senior partners of large and influential New York and Washington law firms.
  • In education, we identify sixty- two colleges and universities with endowment funds totaling $1 billion or more. These institutions control well over two-thirds of all endowment funds in higher education, and they are consistently ranked among the nation’s most “prestigious” colleges and universities. Our leadership group includes their presidents and trustees. We do not include public universities.
  • In the governmental sector, we identify as positions of authority in the executive branch the President and Vice- President, senior White House presidential advisers, and secretaries of all executive departments. In Congress we include the House and Senate majority and minority leaders and party whips, as well as all congressional committee chairpersons. Finally, in the Supreme Court, we include all nine Justices.

When analyzing the education of the elites, Dye found that these elites overwhelmingly came from a few elite universities. In fact, about 50% of the elites received degrees from just 12 universities: Harvard, Yale, Princeton, Columbia, Pennsylvania, Stanford, Chicago, Berkeley, Johns Hopkins, MIT, Cornell, and Northwestern (page 180). Over 20 percent of elites received a degree from Harvard University.

The graduates of elite universities also have significant influence on broader academia. This is demonstrated by the fact that the vast majority of professors in the United States received their doctorates from a tiny number of universities. For example, consider a study by Clauset et al. (2015) [archived] who analyzed faculty hiring of 19,000 tenure-track or tenured faculty in North America in computer science, business, and history. Researchers estimated institutional prestige by analyzing the observed faculty hiring network (i.e. “who hires whose graduates as faculty”). The data indicated that faculty production was highly skewed, “with only 25% of institutions producing 71 to 86% of all tenure-track faculty”. In fact, the supplementary materials [archived] indicate that just 18 schools produced half of all computer science faculty, 16 schools produced half of all business faculty, and 8 schools produced half of all history faculty (Table S2).

Oprisko (2012) [archived] presented data showing that hiring is highly influenced by prestigious institutions among political science faculty. In his book, Honor: A Phenomenology, he compiled data of “the tenure-track and tenured faculty in all ranked research universities to determine which of those universities successfully placed candidates at peer institutions”. The research indicated that just four schools — Harvard, Princeton, Stanford, and the University of Michigan — contribute “roughly twenty percent of the total tenure-track lines in the discipline at research-intensive programs.” Furthermore, the data shows that just “eleven schools contribute 50 percent of the political science academics to research-intensive universities in the United States”.

The graduates of elite universities have tremendous political, economic, and cultural influence on the country. Thus, the admissions policies of these institutions will have tremendous social implications on the rest of the country. There are at least two reasons for this:

  1. The admissions policies of these institutions likely influence who will acquire tremendous social influence in the country. This is assuming that the power of the graduates of elite universities is at least partially caused by attending such universities, rather than it purely being the result of these universities selecting for individuals who will be influential.
  2. More importantly (I believe), the admissions policies of these elite institutions reflect a broader political vision of elite universities. Thus, these admissions policies reflect a broader political vision of the intellectual elite of our country. In this case, these large racial preferences demonstrate the extreme value that our intellectual elites place on vague, undefined appeals to racial “diversity”, to the exclusion of other virtues such as meritocracy, procedural equality, socioeconomic diversity, or viewpoint diversity. If we assume that the political vision accepted by these universities have an influence on the political attitudes of its students, then we can expect our future elites to adopt similar values when they acquire power.

Because of these reasons, I believe we should be very interested in the admissions policies of the elite institutions of higher learning. In particular, we should be concerned about the extreme racial preferences exhibited by our top universities that I’ve documented earlier. We should be actively demanding transparent and clear answers from these elite institutions to the questions that I posed earlier (regarding timeframes, defining “critical mass”, socioeconomic diversity, etc.). Currently, there seems to be no adequate answers to these questions. That should be concerning to us all, since the admissions policies of these institutions affect us all.

4 comments on Racial preferences at elite American universities

  1. Wow. Remarkable effort. Came across a minor typo you’ll want to correct: In the sentence, “That being said, we ought to separate separate Jewish and non-Jewish whites in these statistics…”, the word ‘separate’ is duplicated.
    Thank you for this great resource!

Leave a Reply