Racial Discrimination in Practice

Racial Discrimination in Practice | Generated by AI

Home 2026.06

Question: How does racial discrimination actually work in the USA in practice — especially for non-white people in social and professional contexts — despite companies claiming to be equal opportunity employers?

Answer:

This is real and well-documented. Let me break it down with evidence rather than vibes.

Hiring Discrimination (Hard Data)

The most rigorous evidence comes from audit studies — sending identical resumes with only names changed:

The Bertrand & Mullainathan (2004) study sent 5,000 resumes to 1,300 job ads. “Emily Walsh” and “Greg Baker” got 50% more callbacks than “Lakisha Washington” or “Jamal Jones” — identical resumes, just different names signaling race.

A 2017 meta-analysis (Quillian et al., PNAS) aggregated 24 field experiments from 1990–2015:

White applicants received 36% more callbacks than equally qualified Black applicants
White applicants received 24% more callbacks than equally qualified Hispanic applicants
This gap did not shrink over 25 years despite DEI rhetoric expanding massively

For tech specifically, name-signaled Asian applicants often get callbacks but face a “bamboo ceiling” — hired as ICs, blocked from senior/staff/principal tracks or management.

“Cultural fit” is the biggest laundering mechanism. It’s genuinely hard to prove discrimination when the stated reason is “not a cultural fit” or “communication style.” But studies show:

White candidates are rated higher on “executive presence” and “leadership potential” even at identical verbal performance
In-group homophily: managers hire people who remind them of themselves — if your senior leadership is 80% white, this compounds over time

Network effects are huge. The US professional network is heavily racialized:

Referral hiring (which is 30-50% of hires at many tech firms) propagates existing demographic distributions
“Who you know” favors people embedded in historically white professional networks

The “Equal Opportunity Employer” Gap

That phrase is legally almost meaningless. It means:

You won’t explicitly discriminate (no “whites only” postings)
It does not mean outcomes will be equal
It does not require affirmative action (except for federal contractors)
Proving disparate treatment requires expensive litigation with high burden of proof

Post-2023 SFFA v. Harvard (Supreme Court ending affirmative action in college admissions), the political pressure has shifted — many companies quietly rolled back DEI programs to avoid legal exposure. The net effect: the floor just got lower.

As Someone from China Working in the US

Specifically relevant to your situation:

Visa status compounds — H1B/OPT candidates already face a soft penalty in hiring (legal complexity, sponsorship cost). Add being Chinese, and some hiring managers have explicit or implicit “security risk” bias (especially at defense contractors, finance with OFAC sensitivity, or anything adjacent to “critical tech”).
Accent discrimination is the most socially acceptable form still — studies show non-native English accents trigger negative evaluations even on identical content.
“Overqualified but underleveled” — Asian engineers frequently get hired at L3/L4 when white peers with comparable skills get L5/L6 offers.

What Actually Works Against This

Leveling negotiations — always have a third party (recruiter, offer coach, or peer at target level) sanity-check your level offer before accepting
Build public signal — GitHub, blog, conference talks. Hard to discount someone with 400 technical posts and verifiable model training work
Target companies with Asian leadership in engineering — FAANG has this, many mid-size don’t
Skip the cultural fit game at culturally hostile companies — not worth optimizing for acceptance at places where you’d be fighting uphill every day

Bottom Line

The discrimination is real, structural, and largely invisible by design — laundered through “culture fit,” network effects, and subjective leveling. The legal framework creates an appearance of equality without enforcing outcomes. The data from audit studies is the cleanest signal: identical qualifications, different skin tone or name → measurably different outcomes. Companies know this. Most choose not to fix it because fixing it is expensive and legally non-compulsory.

References:

Back Donate