For some reason, I trust my followers a little too much. When asked a poll of a couple topics I could write about, they inevitably chose this one. In fact, I did a small writeup on this about a year ago as a thread that I’ve now lost. Still, there’s no better time than now to revisit this analysis. We’re just at the end of the academic year for most schools and by now, most universities have sent out their results. People are elated, furious, and confused as with many things in life. I’ve said before that elite college admissions is a flawed, but pretty workable, probabilistic filter for institutional interests. Top colleges want people who are smart, creative, exceptional, newsworthy, future-donators, and hardship cases (to show off to ones before).
I don’t consider myself to really be any of these. I just got lucky in transfer acceptances after my parents made me change schools. A long ramble I wrote a year ago can be found here that mentions this. But, many are unlucky in this rat race. It’s the ideal of the upper-middle class, the 9.9%, and tiger parents who just want what’s best for their children. No shortage of books, papers, blog posts, podcasts, and tweets have been put out about the topic of college admissions. People who desperately want to get in are anxious about their chances. People who get rejected either mope or move on.
After the groundbreaking SFFA v. Harvard case eliminated affirmative action for all universities except military academies, a lot was shaken up in higher education. Universities scrambled to respond and update their admissions criteria as later applicants still complained about inequality or discrimination. Stanley Zhong, who applied in the 2023 cycle, is back in headlines after his initial media circus right after his rejections. Zach Yadegari also gave Twitter a scare earlier this year when he tweeted about his admissions results. Twitter’s conclusion? Elite universities don’t like founders! Ignore the Zuckerbergs, the Elon Musks, and Sam Altmans of the world! College doesn’t matter, they said.
The internet has taken a liking to these two cases and the Trump administration’s crusade against universities isn’t helping. Even after it’s been made illegal, there’s still a flurry of accusations about DEI and affirmative action. Zhong’s case is specifically an accusation that the University of California system racially discriminates in its application process, despite Prop 209 eliminating affirmative action for California public schools in 1996. Well, at least this is somewhat of a transition to Berkeley.
Cal is attractive to analyze for a few reasons. First, it’s a public school. Most immediately, that means they publish their data. Everything found here is taken straight from them. Further as a public school, we carry the expectation that schools subsidized by taxpayers are meant to serve the taxpayers: public schools should serve the public. This is certainly reflected in tuition. California residents pay just under $15,000 in tuition per year while out-of-state and international students pay around $50,000. Also as a public school, it often is beholden to the public through governance. If there’s any school to pressure into implementing systematic changes, it kind of has to be public to avoid the whole lawsuits thing.
Second, it’s an elite undergraduate institution highly ranked in several categories. It’s arguably the best public university (though US News will have you believe its second to its cousin, UCLA) and ranked #17 at the time of writing. Almost any list ranking universities in general puts Cal near the top. They have world-famous departments in EECS, English, math, CS, chemistry, political science, history, earth science… The list is about 15 more majors long.
Last, Berkeley is massive. There are over 30,000 undergraduates at Berkeley, far larger than its private peers. The most popular majors are CS (800/class), economics (700), and cellular/molecular biology (700) just to name a few. That’s both a large sample size and a challenge for admissions in determining where to cut off.
The Numbers
In this cycle, there were a total of 124,242 applicants to UC Berkeley. The overall acceptance rate was around 11% and its yield rate was 46%. The “Others” in the chart include American Indians (0.5%), Pacific Islanders (0.2%), and people who chose not to disclose their race when applying (3.9%). Internationals make up a surprising chunk of applications, nearly 1 in 5.
Moving on to something more interesting, how do the acceptance rates look by race?
Sorted by highest acceptance rate to lowest, American Indians, the unknowns, and Pacific Islanders are pulling their weight. At the lowest, and to be expected, are international students with a 3.3% acceptance rate. Right above them, Blacks are at a 9.9% acceptance rate. This seems to align with most people’s senses. Without affirmative action, Asian people have a higher acceptance rate than most other groups. At Cal, that translates to a 13.1% acceptance rate or about 1 in 8 applicants being accepted.
As for the in-state vs. out-of-state vs. international comparison, in-state students twice as likely than out-of-state students and five times as likely than internationals to be accepted.
Schools!
Let’s move on to a more interesting discussion. Is there extensive school-based and/or geographical inequality going on? The phenomena of a feeder school in this context refers to a school that sends a disproportionate amount of its student body to either particular schools or elite schools in general. Typically, they’re well funded either as private schools or magnet schools. Even as a “regular” public schools, they achieve moderate amounts of funding by being located in an area with wealthy families and expensive homes as part of property taxes go towards public schools. With that money, they provide academic resources, dozens or even hundreds of after-school clubs, and a cutthroat academic environment. That last part leads to an unsettling inference: trauma and mental illness are probably rampant, masked by academic validation. The Atlantic has an solid piece about that issue from 2015.
Looking at this admissions data can tell us two main things (that I’m interested about, anyways). First, what are the feeder schools? Second, are there geographical hotspots? On the second question, the natural inclination is ‘yes’. Surely, the Bay Area and LA are full of private schools and cracked high schoolers? We’ll soon find out.
Again, this matches my general intuition. Private schools have a higher acceptance rate than public schools, in-state schools than out-of-state schools, and out-of-state-schools than international schools. Nothing immediately alarming. If you want your kid to have the best shot at Berkeley, move to California and get them in a private school.
Now for the part that’s the most painful for me: going school-by-school through UC’s horrendously formatted spreadsheet and making many pivot tables. Small caveat here as well. For an unknown reason, UC will list a high school and then not provide a count if the relevant statistic is less than 3. Annoying for a full picture, but the conclusions remain similar enough.
The following is a table of the top 10 schools from each school category (CA public, CA private, domestic, and international) in terms of how many students were admitted. I’ve also added other additional details about the school as I see fit. For tuition information, I’ve opted for the boarding price where possible, have not included any added fees or deposits, and converted any international currencies to a USD estimate. For class size, I’ve used the most recent data from the NCES or sourced from the school itself. GPA data is collected from the UC Information Center when available.
If you attend an elite university or spend your weekends stalking LinkedIn, many of the schools should be familiar. Stanley Zhong’s high school appears 4th on the list. PALY, mentioned in the earlier Atlantic piece and also featured in Ro Khanna’s attack on de-laning, is at 10th. Some honorable mentions that didn’t make the top 10 on the public high schools list include Monta Vista, Lynbrook, Mission San Jose, and Canyon Crest Academy.
The rest of the list also has fairly expected names. Again, props to Loyola, BASIS Silicon Valley, Sacred Heart Prep, and Stanford online as California private schools. The rest of the states looks a little more interesting, featuring classic Northeast boarding schools like Andover and St. Pauls, New York City magnet schools, and some unexpected but highly ranked publics like Lambert, Coppell, and Evanston. Some of the BASIS schools in Arizona, Princeton High School, and a couple of DMV schools get a shoutout as well. As for the internationals sections, the data was a little corrupted with some countries just taking up their own entire row in combination with schools having their own row (seemingly at random). But away from that, primarily being schools in East Asia also makes sense. Further highlights below the cut are Hong Kong International, the Singapore American School, and Dhirubhai Ambani International School.
The following is completely unscientific and not statistically valid, but it does look interesting.
Well, isn’t that curious. In total, there are 9,698 total high schools represented in the UC’s dataset. 1,559 California public schools, 437 California private schools, 5,496 domestic schools, and 2,206 international schools (some variance on international schools because of faulty data). In a combined pivot table, the top 100 schools by admits make up 24.2% of all admits to the school. Roughly, that’s about the top 1% of schools representing 1 in 4 acceptances.
It takes only 354 schools or 3.7% to reach half of the acceptances. To get 80% of the acceptances, it shoots up to 1,185 schools. Even then, 12% of schools accounting for 80% of the acceptances to Berkeley seems like an impressive count. There’s still some recognizable names near that limit if you’re an addict to this like me: Los Alamos High School, Sidwell Friends School, Groton, Xavier, and West Windsor-Plainsboro to name a few.
It’s quite a daunting fact to realize how concentrated this distribution is.
There are 8,451 schools not represented in this distribution because they have 2 or fewer admits to Berkeley. A small fraction of high schools account for the vast majority of acceptances. The steeper curve near the beginning shows that the first few hundred alone make up a massive portion. The opportunity to attend Berkeley seems fairly impacted by your high school, a choice that most students don’t have.
Geography
Because UC’s dataset isn’t as granular when it comes to schools, this part is going to have a few gaps. Because the numbers are per school and include applications, admits, and enrollees for each school, anything below 3 is just an empty cell. Say 10 people apply in a county and 2 people get in. The data would be blank on acceptances and show 10 people applied. Even worse, say there’s only a few schools and each has just 2 applications. The entire county ends up blank. In terms of percent difference, overall applications are 1.5% off and admits are 7.2% off. For smaller counties and schools, the missing 2 admits add up over the 1,996 California schools in the dataset.
Primarily, it’s the superior and central counties that have the most gaps. Inyo only gets a 30% acceptance rate because of one school on record with 3 admits out of 10 applications that year. Kings County also finds itself missing with 6 schools that have blank admit cells. Assuming the best, it could have anywhere from a 12-18% “true” acceptance rate.
Still, there are a couple of interesting observations. LA and the Bay Area are certainly powerhouses. California’s population in 2020 was 39.5 million people and the 9-county Bay Area was 7.76 million; around 19.6%. In terms of applications to UC Berkeley, the Bay comprises 33.3% and a slight overrepresentation at 34.04% of the acceptances. However, the Bay Area acceptance rate is 14.35%, just under the in-state acceptance rate of 14.87%. Generally, it doesn’t seem like the Bay Area is particularly discriminated against and its proportional acceptances ~ admits confer its standing as where almost everyone is form.
One last set of graphs, I promise. Income data comes from the NIH’s HDPulse platform and housing data comes from the California Association of Realtors.
Again, some pockets of missing data led to some counties not being featured in the graphs. I chose a quadratic trend line because that aligns with the intuition (and data) that low income households have a higher acceptance rate than middle income households all else being equal.
Bay Area counties standout again as very high income and housing price locations. That ties in pretty smoothly to school funding and the amount of kids applying to Cal. Given the influence of the tech sector and Berkeley’s EECS program, again, covered many times over. Counties with lower median socioeconomic status still fair pretty strongly in the admissions process. That’s despite both class rank and geographical residency, both factors that could help the applications of students in more low-income districts, not being considered according to Cal’s CDS.
The End
I have a bunch more thoughts on interpretations of these graphs and what they mean for California’s education policy, but those will have to wait another day. I hope you can find some use out of these charts and I’ll do my best to try and make my entire spreadsheet available (if I can clean it up, that is).
Addendum:
I want to thank @anjali_shriva on X for pointing something out. In my combined pivot table with 9698 total schools, the top 97 schools by admits produce 3,241 total acceptances out of 13,639 acceptances. So, the top 1% of schools are responsible for 23.76% of acceptances at Berkeley.
In terms of school concentration, Anjali pointed out that this may be misleading as it doesn’t differentiate between the four categories of schools that the UCs separate by. So, I went back to the pivot tables and got some more numbers.
There are 1,559 California public schools in the data set. The top 1% or 16 schools have 845 admits out of the total CA public count of 8,967. That’s 1% of schools responsible for 9.4% of admits. To reach the 50% of admit mark, you need 180 schools or about 11.6% of all schools.
For California private schools, there are 437 total. The top 1% of schools produce 135 admits (from just 4 schools) and out of the 1,682 CA private acceptances, that’s about 8%. Still overrepresented, but a little less so. To get to 50% of acceptances mark, that takes 43 schools or 9.84% of private schools.
Next, there are 5,498 domestic schools listed for 2,177 acceptances. The top 1% of schools produce 16.2% of the admits. The specific data cuts off quickly, so we’re without the 50% threshold like before. The best I can produce is that 3.29% of domestic schools are responsible for 35.6% of domestic acceptances.
Lastly, with 2,206 international schools, the top 1% get you to 20.3% of acceptances and like the cutoff before, the top 2.77% of schools make up 38% of the acceptances.
The reason for why the overall count differs is because the addition of all these schools makes the “1% of all schools” number larger. Returning to the combined PT with everything, it might be more accurate to say that the top 100 schools produce 24.2% of admits and the top 354 schools make up half of the admits. In the combined table, the top domestic school, Andover, doesn’t appear until the 216th row. In-state high schools are really pulling their weight. The reason for the concentration difference is just that. Still, I don’t think it’s inaccurate to say that the top 1% of schools makeup almost a quarter of acceptances. Either metric or pivot table you go by, that’s still a lot of concentration in a small percentage of schools.