For those who disagree regarding the Harvard case, let me highlight one point. Harvard claimed they found no evidence that their admissions disadvantaged Asian applicants. However, Harvard contended that without affirmative action, more Asians would be admitted resulting in a less diverse student body. This logical inconsistency essentially undermined Harvard's own case. This is why Harvard shifted public opinion away from the idea that they discriminated against Asian applicants and instead towards the argument that reducing affirmative action would lead to a less diverse student body.
The reality is that most of us, especially in California, aren't significantly affected by Harvard or Ivy League admissions. In California, where annual tuition at the nation's competitive universities ranges from $50k to $65k, many hope to take advantage of in-state tuition at the University of California institutions, with Berkeley being the flagship campus.
On a related note, on a school-to-school basis, UCLA is actually more challenging to gain admission to for California Asian applicants. A student from a top international school in Hong Kong or Beijing has a better chance than a student from Whitney High School, Lowell, Palo Alto, etc. However, those same students have a better chance of getting into UC Berkeley than these international students.
This brings us to the central point of this blog post: there are "fixed systems" in place in UC Berkeley Admissions when it comes to Asian applicants. Anecdotally, this has been recognized for a long time, but nowadays, the evidence supports this notion.
I employed a straightforward methodology, gathering data from the University of California website on Asian admits to UC Berkeley for 2022. I excluded data for schools with zero admits. My initial intention was to assess whether schools like Dougherty Valley High School in San Ramon, CA or Mission San Jose in Fremont, CA lived up to their reputations. This led me on a three-day data mining and machine learning endeavor, uncovering what I believe is a startling revelation about admission to UC Berkeley for Asian applicants.
Regarding the methodology, I sensed that the story was incomplete using only the University of California website data. I decided to supplement it with information from each school's School Accountability Report Card (SARC). I manually input the 12th-grade student body size, the number of Asian students in that class, and the percentage of economically disadvantaged students attending the school as a whole. It would be more convenient if the California Department of Education compiled this information into a single file rather than scattered PDFs across the internet.
Instead of solely examining the number of Asian applicants versus admitted students at UCB, I analyzed the number of Asian students in the 12th grade and how many of them were admitted. This is where I discovered the implementation of the "system."
Using a simple Pearson's Correlation, the 2022 data exhibited a significant negative correlation of -0.3815 with a P-Value of 4.1466e-12. In essence, this indicates that as the percentage of Asian students in the 12th-grade class increases, the acceptance rate decreases and vice versa. Simply put, an Asian applicant has a better chance of being admitted to UC Berkeley if they come from a school with fewer Asian students.
I also examined the relationship between the total number of students in the 12th-grade class and the acceptance of Asian students. This investigation revealed a noteworthy finding: a meaningful negative connection with a value of -0.3284. This essentially indicates that as the size of the 12th-grade class grows, the rate of acceptance for Asian students tends to decrease. The statistical evidence for this is quite strong, as indicated by a p-value of 3.5192e-09, which signifies that this relationship is not likely due to random chance.
When examining the correlation against the percentage of economically disadvantaged students, no significance was observed. At least on the surface, it seems that if you're an Asian applicant from an economically disadvantaged neighborhood, it doesn't impact your application. This is intriguing, as one might assume that income would be a factor if UC Berkeley aims to diversify its student body, but it doesn't appear to be the case for Asians.
I then applied several basic machine learning models: LightGBM, CatBoost, Random Forest, Ridge, Lasso, SVR, Gradient Boosting, and a simple Neural Network Architecture. I divided the approximately 300 entries into test, train, and validation sets in a 70:15:15 ratio.
The best outcome was achieved with CatBoost, yielding a Mean Squared Error of 0.0029 and an R-squared value of 0.4094.
In the context of mean squared error, a smaller value is preferred, indicating accurate predictions close to the actual values. The mean squared error value of 0.0029 is extremely small, implying the model's predictions are highly accurate.
A higher R-squared value is desirable, and 0.409 is quite good. This suggests the CatBoost model effectively explains around 41% of the reasons behind variations in the numbers of Asian students admitted across different schools.
A 41% explanatory power is significant, especially considering that this factor lies beyond students' control, unlike variables such as grades and extracurricular activities. It's important to note that an Asian student can't reasonably compel their parents to relocate to a school with fewer Asian students to improve their chances.
Continuing my investigation, I focused on schools with higher-than-average Asian applicants, which is 62.3. I included data on how many Asian students met or exceeded assessments in English Language, Math, and Science, along with the results for the entire student body. This limited the study to cover 93 schools, encompassing both elite institutions like Whitney, La Canada, San Marino, Mission San Jose, Dougherty, Lynbrook, Palo Alto, and others with substantial Asian populations that aren't typically considered "elite."
Regardless of whether I assessed the student body assessment scores, Asian-only assessment scores, or the differences between them, no statistical significance emerged. Essentially, it seems that the competitiveness of your school doesn't significantly impact an Asian applicant's chances of being accepted into UC Berkeley. At least for 2022, there's no evident advantage to attending a highly competitive public high school in California, contrary to common perception.
For example, being in the top 10th percentile at Lynbrook High in San Jose, where 96% of students met or exceeded English assessment levels, versus Abraham Lincoln in San Francisco, where only 55% of students met or exceeded the same standard, appears to have no discernible influence.
That's reassuring for some but bewildering for others. For those who clearly don't rank within the top 10 or 15 percentile of their class at an elite public school, this situation creates a challenge since at a less competitive school they could certainly be ranked higher, though not necessarily. After all, Asian students at less competitive high schools might possess intelligence equivalent to the highest achievers at competitive schools. Yet, my current data lacks the capacity to ascertain this.
However, the discovery is disconcerting as it appears to involve a form of racial balancing, likely obscured by some type of geography. So far, it seems that the size of the 12th-grade student body and the percentage of Asian students in that class can account for up to 40% of the variation in the number of Asian students accepted from a specific high school into UC Berkeley. The school's competitiveness and its location in a low-income neighborhood have no impact on Asian applicants. This contradicts the idea that the UC system's application process is holistic, particularly when student body size and composition are beyond a student's control.
I want to acknowledge the limitations of this study. It only examines data from 2022 and lacks a trend analysis. Furthermore, the data is derived from approximately 300 high schools in California. While I am not a data scientist, I hold degrees from UC Berkeley's Haas School of Business and a Master's in Finance from London Business School. This is to assure that I am not simply a conspiracy theorist or a right-winger crying wolf. Though I am not an expert in machine learning, I've spent around four years studying it for financial purposes. In comparison, other subjects are much simpler than the complexities of financial time series analysis. I've gained enough knowledge to challenge a so-called machine learning expert and their course on financial time series. It's important to note that the intricacies of financial time series, like forward-looking bias, overfitting, and temporal dynamics, make it a challenging domain.
As a parent of an Asian toddler who will likely grow up in a low-income household, I still align with the Democratic party in my voting. However, due to my analysis of data, I now consider myself more of an independent. This perspective arises from the realization that "liberal" media, like "conservative" media, often has an agenda. It would be naive not to acknowledge this. While I agree that diversity is essential, I believe in transparency in the methods employed to achieve it. Discrediting an entire population by propagating fabricated issues or altering the rules once Asian students excel is unjust.
An illustrative example is evident in the Harvard admission process. Despite Asian students scoring equally to other applicants in "personality" during alumni interviews, the admissions department unjustly marked them down. Conversely, they inflated the scores of wealthier students. Curiously, people reacted more when they heard about the inflation of scores for affluent students, but largely ignored the devaluation of Asian students to make space for others.
For Asian students and parents, I offer words of encouragement. Recognizing this as a statistical reality (I encourage others to replicate my findings and cite me), it's important to understand that this might not significantly affect you. What do I mean? UC Berkeley and UCLA confidently reject qualified students because they know these students will likely secure admission in other reputable universities. This is perhaps their way of diversifying the distribution of Asian students.
Another interesting study, conducted ironically by a Harvard Professor, is called the Opportunity Atlas. It examines economic mobility over time for low-income, middle-income, and high-income individuals across various racial backgrounds, including Black, White, Hispanic, Asian, and Native American. According to this project, a Black male earning an annual income of $27,000 at age 35 is considered upwardly mobile, whereas an Asian male from a low-income family with the same income would be seen as below average. It's clear that Black Americans face significant economic disadvantages, and it would be hypocritical to ignore this fact.
The truth is that most Asian students will likely fare well. It's a testament to Asian Americans that these esteemed institutions are reshaping admission rules due to the outstanding performance of a small minority. Despite comprising only 7.3% of the US population, Asian Americans have prompted institutions like Harvard and UC Berkeley to enact measures to curb what could have been an overwhelming majority.
This reminds me of my youth when America attempted to convince everyone that Asians lacked leadership skills, creativity, and other attributes. Over time, I've realized the falsehood of these claims. They were mere racist rhetoric intended to provide subjective justifications for biased decisions. This parallels how Harvard unjustly marked down Asian students' "personality" without even meeting them.
Nevertheless, although our children are likely to thrive in comparison to their non-Asian peers, it doesn't justify misleading an entire population into believing that biases against Asians don't exist. It's equally wrong to prioritize the struggles of other minorities over those of an ethnic group that has also experienced historical disadvantages.
In practical terms, here's my strategic advice for Asian high school students. If possible, apply to as many universities as you can. While I initially believed that five applications were sufficient, I was surprised to learn that a non-Asian student who got into Yale applied to ten universities. This might be financially burdensome, especially for low-income households, but if feasible, it offers the best chance of securing admission to a top 25 university—provided you are a committed student aiming for excellence.
For parents considering moving for a "school district," prioritize safety and community over the district's reputation. When a school district becomes popular among Asians, you're automatically at a disadvantage. The algorithms governing these dynamics appear to be in constant flux, making it challenging to predict the landscape five years ahead. Financially, consider “value” cities, for example, the cheapest home in Mission San Jose School district is $1.78MM, but it’s below the average of acceptance for the 12th Grade Asian Student Body Size. Meanwhile, in Lafayette, which is currently above average for acceptance rate, has three properties in the $1.2mm-1.3mm range. If you’re wealthy enough to be buying houses that expensive, then chances are your financial aid package will be small to non-existent so you better save for the actual tuition for whatever school your child ends up going.