Source: hidden-rules.md
Elite college admissions operate on a fundamentally different set of rules than the meritocratic process most families imagine. Data from the SFFA v. Harvard lawsuit (2019-2023) revealed, for the first time, the internal mechanics of an elite admissions office -- exposing how hooks, institutional priorities, and strategic timing dramatically reshape who gets in. This document synthesizes real data from that lawsuit, admissions office disclosures, and expert analyses to identify the hidden variables that drive admissions outcomes.
ALDC = Athletes (recruited), Legacies, Dean's/Director's Interest List (donors), Children of faculty/staff.
Data from the Arcidiacono, Kinsler & Ransom analysis of Harvard admissions records (six admission cycles, Classes of 2014-2019):
| Hook Category | Admit Rate | Comparison to Unhooked (~5-6%) |
|---|---|---|
| Recruited Athletes | 86% | ~15x unhooked rate |
| Children of Faculty/Staff | 47% | ~8x unhooked rate |
| Dean's/Director's Interest List (donors) | 42% | ~7x unhooked rate |
| Legacy Applicants | 33% | ~6x unhooked rate |
| No ALDC Hook (unhooked) | ~5-6% | baseline |
Source: "Legacy and Athlete Preferences at Harvard" (Arcidiacono, Kinsler, Ransom -- NBER Working Paper 26316, published in Journal of Labor Economics 2022)
43% of white admits at Harvard are ALDC
< 16% of African American, Hispanic, and Asian American admits are ALDC
Roughly three-quarters of white ALDC admits would have been rejected if treated as typical white applicants
Removing ALDC preferences would cause the share of white admits to fall and all other groups to rise or remain unchanged
Recruited Athletes (biggest hook):
Coaches provide a ranked list to admissions; top recruits get "likely letters" (near-guaranteed admission)
The coach's support effectively acts as a near-automatic admit for top-priority recruits
Athletes are evaluated on a lower academic bar (must meet a floor, but the floor is well below the typical admit)
Legacy (children of alumni):
A white applicant with a typical 10% chance sees a 5x increase in admission likelihood as a legacy
Primary legacy (parent attended) is strongest; secondary legacy (grandparent, sibling) is weaker
Some schools (e.g., MIT, Caltech, Johns Hopkins post-2020) do not consider legacy status
Dean's/Director's Interest List (donors):
Contains applicants "of special importance to the dean of admissions" -- primarily children of major donors or those with donation potential
A white applicant with a typical 10% chance sees a 7x increase if on the dean's list
Not publicly disclosed; managed through development office communications with admissions
Children of Faculty/Staff:
Smaller category but significant boost (47% admit rate at Harvard)
Applies to full-time faculty and sometimes senior staff
Students can have multiple hooks (e.g., legacy + recruited athlete)
Hooks generally stack, though the marginal benefit diminishes: a recruited athlete who is also a legacy already has near-certain admission
The most powerful combination is recruited athlete + any other hook = near-100% admission probability
| Hook | Recommended Multiplier (on base admit probability) |
|---|---|
| Recruited Athlete | 3.5-4.0x (capped: if base > 25%, cap at ~90%) |
| Donor/Dean's List | 3.0-3.5x |
| Legacy | 2.5-3.0x |
| Faculty/Staff Child | 2.0-2.5x |
| First-Generation | 1.3-1.5x |
| Multiple hooks | Multiplicative but with diminishing returns (apply largest hook fully, additional hooks at 50% bonus) |
Universities:
| School | ED Rate | RD/Overall Rate | ED Multiplier |
|---|---|---|---|
| Brown University | 14.4% | 5.4% overall | 2.7x |
| Columbia University | 13.2% | 3.9% overall | 3.4x |
| Cornell University | 17.5% | ~8% RD | 2.2x |
| Dartmouth College | 19.1% | 5.4% overall | 3.5x |
| Duke University | 19.7% | 6.7% overall | 2.9x |
| Emory University | 23.2% | 10.2% overall | 2.3x |
| Johns Hopkins | ~14% | ~8% overall | 1.8x |
| Northwestern University | 23% | 7.7% overall | 3.0x |
| Rice University | 16.8% | 7.9% overall | 2.1x |
| UPenn | 14.2% | 5.4% overall | 2.6x |
| Vanderbilt | 13.2% | ~6% overall | 2.2x |
| WashU (St. Louis) | 25.2% | 12% overall | 2.1x |
Liberal Arts Colleges:
| School | ED Rate | Overall Rate | ED Multiplier |
|---|---|---|---|
| Amherst College | 29.3% | 9% | 3.3x |
| Middlebury College | 30.5% | 10.7% | 2.9x |
| Williams College | 23.3% | 8.3% | 2.8x |
| Wellesley College | 29.8% | 14% | 2.1x |
| Barnard College | 25.6% | 8.8% | 2.9x |
EA Schools (non-binding):
| School | EA Rate | Overall Rate | EA Multiplier |
|---|---|---|---|
| Harvard | ~9% | 3.6% | 2.5x |
| Yale (SCEA) | 10.8% | 4.5% | 2.4x |
| MIT | 5.2% | 4.5% | 1.2x |
| Georgetown | ~15% | 12.9% | 1.2x |
| Notre Dame | 12.9% | 11.2% | 1.2x |
Sources: CollegeVine, Spark Admissions, College Kickstart, IvyWise (Class of 2029/2030 data)
Students who need to compare financial aid packages across schools (ED is binding before you see aid offers from other institutions)
Students uncertain about their top choice
This creates a socioeconomic bias: wealthier families who don't need to compare aid can commit ED; lower-income families often cannot
Some schools (e.g., QuestBridge partners) offer ED with guaranteed need-met aid to partially mitigate this
EDII is a second binding early round with a January deadline (results in mid-February), used by students who were deferred/rejected from ED I at another school.
Schools offering EDII include: Vanderbilt, WashU, Emory, Tufts, Middlebury, Bowdoin, Pomona, Claremont McKenna, Colby, Wellesley, Brandeis, NYU, Boston College, Boston University, Lehigh, Case Western, and others.
EDII boost is real but smaller than EDI:
The class is partially filled from ED I, leaving fewer spots
Typical EDII multiplier: 1.3-1.8x vs. RD (compared to 2-3x for EDI)
Example: Middlebury ED 30.5% vs. 10.7% overall; Johns Hopkins ED ~14% vs. ~8% overall
| Round | Multiplier on Base Admit Probability |
|---|---|
| ED I | 1.5-2.0x (accounts for both the boost and self-selection) |
| EA/REA (non-binding) | 1.1-1.3x (mild signal of interest; smaller pool effect) |
| ED II | 1.3-1.5x |
| RD | 1.0x (baseline) |
Note: These are "net" multipliers for simulation -- lower than raw rate ratios because some of the raw ED advantage comes from applicant pool quality differences, not purely from the binding commitment boost.
Yield protection is the practice of rejecting or waitlisting applicants who are overqualified for a school, on the assumption that they will be admitted to more prestigious institutions and decline the offer. Schools do this to protect their yield rate (% of admitted students who enroll), which factors into rankings and institutional reputation.
No school has ever officially admitted to practicing yield protection
Naviance scattergrams (plotting GPA/test scores vs. admit decisions) sometimes show a distinctive pattern: admit rates increase with stats up to a point, then decrease at the very highest levels -- the "inverted U" pattern
College counselors and admissions consultants report observing this pattern at specific schools
Statistical evidence is anecdotal rather than based on controlled studies
| Frequently Accused | Occasionally Accused |
|---|---|
| Tufts University | University of Michigan |
| Tulane University | UVA |
| Northeastern University | UC campuses (various) |
| Case Western Reserve | Clemson |
| University of Chicago | Auburn |
| Boston University | Colgate |
| Emory University | Lehigh |
| University of Richmond | -- |
Typically triggered when an applicant's stats are significantly above the school's 75th percentile AND the applicant shows no demonstrated interest (no campus visit, no "Why Us" essay, no ED application)
The threshold is roughly: stats > 75th percentile by 100+ SAT points or 0.3+ GPA points, with no demonstrated interest signals
Applying ED or EDII effectively neutralizes yield protection (binding commitment = guaranteed yield)
Writing a compelling "Why Us" essay that shows genuine, specific interest also mitigates the risk
Outright rejection is less common; waitlisting is the more frequent yield-protection outcome
The applicant profile: perfect stats, generic application, no school-specific engagement
Schools that heavily track demonstrated interest are more likely to practice yield protection
For schools ranked roughly 15-40 (not HYPSM/Ivy-tier, but selective enough to care about yield):
If applicant academic index is > 1.5 standard deviations above the school's median AND the applicant did not apply ED/EDII AND showed no demonstrated interest: apply a 0.6-0.8x penalty to base admit probability
If the applicant applied ED: no penalty (yield protection irrelevant)
| Category | Your Admit Probability | Your Stats vs. School's Profile | Typical School Acceptance Rate |
|---|---|---|---|
| Safety | > 70-80% | Above the 75th percentile of admits | Usually > 40-50% |
| Target / Match | 30-70% | Between the 25th and 75th percentile of admits | Usually 25-50% |
| Reach | 10-30% | Below the 25th percentile, OR school has very low acceptance rate | Usually < 25% |
| High Reach | 5-15% | Well below 25th percentile at a highly selective school | Usually < 15% |
| Lottery | < 5-10% (for anyone) | Stats barely matter; outcome is essentially random for unhooked applicants | < 10% overall |
Any school with an acceptance rate under ~10% is effectively a lottery for unhooked applicants
Even perfect-stat applicants (1600 SAT, 4.0 UW GPA, national-level ECs) face < 20% admit rates at HYPSM
The term "lottery" reflects the reality that at < 10% acceptance rates, the variance in outcomes is dominated by factors outside the applicant's control (reader assignment, committee dynamics, institutional needs that year, randomness)
College counselors increasingly tell students: "Every T20 is a reach for everyone. There are no target T20s."
For a student with a given Academic Index (AI) relative to a school's median:
| Student AI vs. School Median | Classification | Base Admit Probability Range |
|---|---|---|
| AI > school 75th + 1 SD | Safety | 70-90% |
| AI between 50th and 75th | Target | 30-60% |
| AI between 25th and 50th | Low Target / Reach | 15-35% |
| AI below 25th | Reach | 5-20% |
| School acceptance rate < 10% | Lottery for all unhooked | cap at school's rate * 1.5 for best applicants |
First-gen status provides a modest but real boost at most selective schools
Schools value first-gen students for socioeconomic diversity and as evidence that their financial aid is reaching underserved populations
Estimated boost: 1.3-1.5x multiplier on base probability (smaller than ALDC hooks, but meaningful)
Many schools have specific programs and recruitment pipelines for first-gen students (QuestBridge, Posse Foundation)
The Supreme Court ruled in June 2023 (SFFA v. Harvard) that race-based affirmative action in admissions violates the Equal Protection Clause
Schools can no longer use race as a direct factor in admissions decisions
However, applicants can still write about how race has affected their life in essays
Schools are shifting toward race-neutral proxies: socioeconomic status, neighborhood disadvantage indices, first-generation status, geographic diversity
Early data (Harvard Class of 2028): 4% decrease in Black enrollment, 2% increase in Hispanic enrollment, no change in Asian American enrollment compared to Class of 2027
For simulation purposes: race should not be a direct admissions factor post-SFFA; instead, model the correlated socioeconomic factors
Students from underrepresented states (Great Plains, Deep South, Rocky Mountain states, rural areas) receive a meaningful admissions boost
Only 9% of Princeton's Class of 2028 came from rural areas, despite 19% of the U.S. population being rural -- schools actively try to recruit from these areas
International applicants face different (often lower) admit rates because they compete in a separate pool and many need financial aid
Estimated boost for underrepresented geography: 1.2-1.4x multiplier
States that are overrepresented (CA, NY, MA, NJ, CT) provide no geographic boost
How it's tracked: campus visits, email opens/clicks, virtual tour attendance, admissions event attendance, alumni interviews, "Why Us" essay quality
Who tracks it: Many mid-tier privates (Tulane, Lehigh, American, GW, Case Western, Northeastern, BC, BU). About 15.7% of colleges rated it "considerably important" per NACAC data.
Who does NOT track it: Ivy League schools, MIT, Stanford, Caltech, most large public universities (they get too many applications to track individual interest)
Demonstrated interest matters most at schools ranked ~20-50 where yield rates are a concern
Estimated boost when demonstrated interest is high: 1.1-1.3x; penalty when absent at schools that track it: 0.7-0.9x
Applying to an oversubscribed major (CS, business, engineering) at schools that admit by major can significantly reduce admit probability
Example: Carnegie Mellon CS admission rate is far lower than CMU's overall rate
Conversely, applying to an undersubscribed program (classics, philosophy, certain sciences) can help
Gender dynamics: engineering/CS applicant pools skew heavily male, so women applying to engineering may have a slight advantage at some schools; the reverse applies for nursing or education programs
For simulation: if a school admits by major, apply a 0.7-0.9x modifier for competitive majors and 1.1-1.3x for less competitive majors
Every year, admissions offices have specific institutional needs that vary:
The orchestra needs an oboist: performing arts departments submit wish lists; if you play an instrument the ensemble needs, you get a meaningful boost (not as large as recruited athlete, but similar to legacy-level)
Residential life considerations: some schools target specific geographic distributions for housing assignments
Gender balance: schools that trend one direction may give a boost to the underrepresented gender
Academic department requests: a department may be trying to grow enrollment, requesting more students interested in their field
These needs are unpredictable and change annually, making them function as additional randomness in the simulation
For simulation: model as a small random "institutional need" bonus (0-15% added probability) applied to a random subset of applicant profiles each cycle
Harvard uses a 1-6 scale (1 = best, 6 = worst) with +/- modifiers across six dimensions:
| Dimension | Weight (relative) | What It Measures |
|---|---|---|
| Academic | Highest | GPA, test scores, rigor of coursework, intellectual curiosity, academic growth potential |
| Extracurricular | High | Depth and impact of activities, leadership, awards, national-level achievement |
| Athletic | Medium | Varsity sports achievement (separate from recruited athlete hook) |
| Personal | High | "Humor, sensitivity, grit, leadership, integrity, helpfulness, courage, kindness" |
| Recommendations | Medium | Teacher and counselor letters; strength of endorsement |
| Alumni Interview | Low | 30-min interview report; limited weight in decisions |
| Rating | Meaning | Approximate Percentile |
|---|---|---|
| 1 | Outstanding / Top nationally | Top ~1% of applicants |
| 1+ | Exceptional, transcendent | Top ~0.1% |
| 2 | Very strong | Top ~5-10% |
| 2- | Strong with minor caveats | Top ~10-15% |
| 3+ | Generally positive, above average | Top ~20-30% |
| 3 | Average for Harvard pool | Middle of applicant pool |
| 4 | Below average / "bland, somewhat negative, or immature" | Bottom half |
| 5-6 | Weak / Very weak | Bottom quartile |
Academic 1 + Personal 1: Near-certain admission (barring institutional constraints)
Overall 2- or better: Advances to full committee review
Overall 3+: First reader decides on case-by-case basis whether to advance
Overall 3 or worse: Typically does not advance to committee
Students scoring a 1 in any major category are almost always accepted
Students scoring 3- or below generally never are
For applicants in the borderline zone (overall ~2- to 3+ range), small positive factors can "tip" the decision:
A compelling personal story or essay
An unusual extracurricular achievement
Geographic diversity (from an underrepresented state)
First-generation status
A specific institutional need (the orchestra needs an oboist)
A strong alumni interview report
Legacy or donor connection (smaller hooks that don't guarantee admission but can tip a borderline case)
The tip factor is what makes elite admissions feel random -- two nearly identical applicants can have different outcomes based on which "tip" the committee values that year.
For the simulation, map the Harvard-style system to a composite score:
| Component | Weight in Composite | Input Variables |
|---|---|---|
| Academic Index | 40% | GPA (normalized) + SAT/ACT (normalized) + courseload rigor |
| Extracurricular Rating | 25% | EC tier (national > state > school-level), leadership depth |
| Personal/Essay Quality | 20% | Random factor simulating essay quality (partially correlated with archetype) |
| Recommendations | 10% | Correlated with academic index + random noise |
| Interview | 5% | Small random factor |
Then apply multipliers for hooks, round, demonstrated interest, and institutional needs on top of the composite score.
| Factor | Multiplier | Notes |
|---|---|---|
| Recruited Athlete | 3.5x | Cap effective probability at ~90% |
| Donor/Dean's List | 3.0x | |
| Legacy | 2.5x | Primary legacy (parent); secondary legacy ~1.5x |
| Faculty/Staff Child | 2.0x | |
| First-Generation | 1.4x | |
| Underrepresented Geography | 1.3x | Rural, Great Plains, Deep South, Rocky Mountain |
| ED I Round | 1.8x | Net of pool quality adjustment |
| ED II Round | 1.4x | |
| EA/REA Round | 1.2x | |
| RD Round | 1.0x | Baseline |
| Demonstrated Interest (high) | 1.2x | Only at schools that track it |
| No Demonstrated Interest | 0.8x | Penalty at schools that track it |
| Yield Protection Penalty | 0.7x | When stats >> school median and no ED/interest |
| Competitive Major | 0.8x | At schools that admit by major |
| Less Competitive Major | 1.2x | At schools that admit by major |
| Institutional Need Match | 1.3x | Random annual assignment |
| Multiple Hooks | Largest hook full, additional at 50% bonus | Diminishing returns |
Arcidiacono, P., Kinsler, J., & Ransom, T. (2022). "Legacy and Athlete Preferences at Harvard." Journal of Labor Economics, 40(1). NBER Working Paper 26316
Spark Admissions: Early Decision and Early Action Acceptance Rates
College Kickstart: Early Decision Schools That Double Admission Odds
College Transitions: Early vs. Regular Decision Admission Rates
Harvard Crimson: "Harvard Ranks Applicants on 'Humor' and 'Grit,' Court Filings Show"
SFFA v. Harvard Findings of Fact and Conclusions of Law (2019)