UC Oceanview — Admissions Impact Simulator

Adverse Impact Analysis Tool

Who Loses Their Seat?

Before you begin. UC Oceanview is a fictional University of California campus built for teaching. The campus, the budget figures, the five administrators, and their dialogue are invented. The legal and policy context is real: Proposition 209 (1996), SFFA v. Harvard (2023), and the UC system's test-blind admissions since 2021.

Your role. You sit where Christine Dao, Vice Provost for Enrollment Management, sits. Five senior administrators each argue for a different identity-neutral rule for deciding which 2,800 admits lose their seats. None of them may adjust for race. Dao decides. Here, you decide.

How to use this page. Read the five proposals, choose one, then look at what it does to the class. The order matters: the case asks students to commit to a criterion before seeing its consequences.

Where the numbers come from. The 8,000-admit pool is simulated: every count on this site comes from parameters set by hand, not from UC records or any other real dataset. What is not invented is the pattern each parameter encodes — AP/IB course availability correlates with school funding and poverty level, how a given GPA translates across schools correlates with the school's average achievement, and specific communities concentrate in specific California zip codes. The numbers make a real structural pattern visible; they are not a measurement of it.

The California legislature has cut UC system funding by 18%. UC Oceanview must reduce its incoming class from 8,000 to 5,200 admits — removing 2,800 applicants from the admit list before offers are sent. Race-conscious adjustments are prohibited under Proposition 209 (1996) and SFFA v. Harvard (2023).

Five senior administrators have proposed identity-neutral criteria for determining which 2,800 admits will be removed. Two of these options — 1a and 1b — use the same construct (academic rigor) but operationalize it differently: one measures absolute AP/IB course count, the other measures courses taken relative to courses available. Select any option below to see its adverse impact on the incoming class.

Admit Pool: 8,000 Admits

By Racialized Identity

By Gender

First-gen 31% · Pell-eligible 34% · Middle-50% SAT 1133–1313 · Mean GPA 3.41 · Mean AP/IB courses 4.7

Select a Removal Criterion

Select an option above
The four-fifths screen (badge above): a legal threshold, not a finding of fairness.

For each group, retention rate = (admits − removed) ÷ admits. The screen divides each group's retention rate by the highest group's rate and flags any ratio below 0.80 (EEOC Uniform Guidelines, 29 C.F.R. § 1607).

Clearing the screen is a legal threshold, not evidence that a criterion distributed its harm evenly. Groups of a few dozen admits can cross it on a handful of students.

Reading the sign: a positive number is bad news.

"Adverse impact" is the harm; the plus sign measures how much of it, not whether it is good. The figure is a gap, in percentage points, between two shares — a group's share of the 2,800 removed, minus its share of the 8,000 admits.

  • Positive — over-removed: the criterion removed more of the group than its share of the pool.
  • Negative — under-removed: it removed less than the group's share of the pool.
  • Zero: removals track the pool exactly — the lottery benchmark on the next page.

Under Option 1a, Hispanic admits are 27.3% of the pool and 42.2% of the removals: +14.8 pp. Asian admits are 27.6% of the pool and 12.4% of the removals: −15.3 pp.

Column definitions: Admits N = total admits in this group. Removed N = number removed under this option. Admit Pool % = group's share of all admits. Removed % = group's share of those removed. Adverse impact (pp) = Removed % minus Admit Pool %; positive = over-removed relative to the group's size in the pool; negative = under-removed. Status = a representation flag only: over-removed above +2 pp, under-removed below −2 pp, otherwise proportional. It is independent of the four-fifths screen (above), and a group can be flagged by one and not the other.
GroupAdmits NRemoved NAdmit Pool %Removed %Adverse impact (pp)removal share − pool shareStatus±2 pp flag, not the legal test

Adverse Impact by Racialized Identity

above zero = over-removedwithin ±2 pp = proportionalbelow zero = under-removed

Adverse Impact by Gender

above zero = over-removedwithin ±2 pp = proportionalbelow zero = under-removed
The Framework

The 2×2: Every Criterion, A Different Harm

Each option occupies one cell in a matrix defined by mechanism (how the criterion produces disparity) and primary harm (which group bears the cost). Options 1a and 2 harm Black and Hispanic admits through different mechanisms. Options 3 and 4 harm Asian admits through different mechanisms. Option 1b uses the same construct as 1a but changes the denominator — shifting who bears the cost.

Disparate Allocation
(structural sorting excludes)		Disparate Valuations
(biased measurement devalues)
Harms Black &
Hispanic admits		 Option 1a: AP/IB Requirement (Absolute)
 Option 2: Unweighted GPA Floor
Harms Asian
admits		 Option 3: Geographic Caps
 Option 4: Holistic Review (HRC)
Shifts harm toward
affluent under-utilizers		 Option 1b: Contextual Rigor
Same construct as 1a — but the denominator changes. Measurement design is an equity intervention.
 This cell remains open — an equity-conscious valuation that reduces disparate impact without introducing identity.
The convergence question: Options 1a and 2 both harm Black and Hispanic admits — through different mechanisms but with the same result. Why do most "neutral" criteria point in the same direction? Because the structural inequality in K–12 education runs in one direction. The criteria are neutral. The structure is not.
The negative action question: Option 4 harms Asian admits — and still clears the four-fifths rule, at a ratio of 0.82 against a 0.80 threshold. The only group the screen flags is American Indian or Alaska Native admits, and at 45 admits that flag reflects small-sample volatility rather than the structural pattern. The bias embedded in holistic review is real but legally invisible. Eliminating affirmative action (Prop 209, 1996) did not eliminate this bias. It has nothing to do with affirmative action. It is negative action (Kang, 1996, 2025).
The denominator question: Options 1a and 1b use the same construct — academic rigor. The only difference is the denominator. Adding school-level context to the denominator changes who is harmed. Under 1a, a student at a rural school with 3 AP courses who took all 3 is removed. Under 1b, that same student has a rigor ratio of 1.0 and is retained. How you operationalize a criterion determines who bears the cost.
Option 5: What if we used a lottery? →