Post-SFFA Demographic Shifts at Selective Colleges

post_sffa_demographics.md

Post-SFFA Demographic Shifts at Selective Colleges

Research compiled 2026-03-09. Covers Year 1 (Class of 2028, entering Fall 2024) and Year 2 (Class of 2029, entering Fall 2025) enrollment data following the June 2023 SFFA v. Harvard Supreme Court decision banning race-conscious admissions.

Executive Summary

Two full admissions cycles after the Supreme Court's SFFA ruling, the demographic effects are larger and more persistent than initially modeled in the simulation. The current SFFA_MULT_BY_TIER table in sim.js significantly underestimates the magnitude of change at Tier 1-3 institutions while approximately capturing the public university countertrend.

Key findings: - Black enrollment at HYPSM: Dropped 36-56% from pre-SFFA baselines (sim models -28%) - Hispanic enrollment at HYPSM: Dropped 14-35% (sim does not differentiate Hispanic from URM) - Asian enrollment at HYPSM: Increased 25-40% (sim models +18%) - Year 2 (C/O 2029) did not recover: Most schools saw continued declines or flat enrollment in the second cycle - Public flagships: URM enrollment increased 8% system-wide (sim models +25-30%, possibly too high) - LACs: Mixed results; some saw dramatic drops (Middlebury -26% SOC), others held steady (Williams) - Data suppression: Many colleges stopped publishing racial demographics in Year 2, making comprehensive tracking harder

Methodology Notes

Data Comparability Challenges

1. Reporting methodology changes: Many schools shifted from self-reported (multiple identities counted) to IPEDS methodology (single category per student) between C/O 2027 and 2028/2029, making direct comparisons unreliable.
2. "Unknown race" inflation: The share of students declining to report race has risen substantially (Princeton: 7.7% to 8.2%; many schools 5-10%), potentially masking true demographic shifts.
3. Data suppression: Fewer than half the schools that released demographic data in Year 1 did so in Year 2. Only 29 schools in the Murphy tracker as of January 2026.
4. International student exclusion: IPEDS excludes international students from racial categories. Schools with large international populations (CMU 38%, Caltech 22%, Columbia 18%) show distorted domestic percentages.

Pre-SFFA Baseline

For consistency, this document uses the Class of 2027 (admitted spring 2023, before the June 2023 ruling) as the pre-SFFA baseline where available, and averages of C/O 2025-2027 where individual years are noisy.

Tier 1: HYPSM (Harvard, Yale, Princeton, Stanford, MIT)

Harvard

*Harvard changed its reporting methodology for C/O 2028, calculating percentages among students who chose to report race rather than the full class. This inflated C/O 2028 minority percentages, making them not directly comparable. The C/O 2029 data uses the same revised methodology, showing the true Year 2 trajectory: continued decline in Black enrollment, further Asian increases.

Sources: Harvard Crimson (Oct 2025), Harvard Magazine, Harvard Gazette.

Yale

Yale's Year 1 data was anomalous: Asian enrollment decreased while Black/Hispanic held steady, contradicting peer trends. Year 2 shows a sharp correction: Asian enrollment rebounded to 30% while Hispanic dropped from 19% to 13% (-32% from C/O 2028, -28% from baseline). This suggests Year 1 reflected application/yield shifts rather than admission policy, with Year 2 revealing the structural effect.

Sources: Yale Daily News (Sep 2025), Yale Office of Undergraduate Admissions class profile.

Princeton

Princeton is remarkable: Year 1 showed almost no change, but Year 2 saw Black enrollment crash to 5%, the lowest since 1968. This delayed effect may reflect: (a) Year 1 pipeline still included students recruited under race-conscious practices, (b) yield effects taking a cycle to propagate, or (c) increased "unknown" masking Year 1 shifts. Asian enrollment also shows a delayed increase.

Sources: The Daily Princetonian (Sep 2025), Princeton Admission Statistics.

Stanford

*Stanford's pre-SFFA Asian percentage depends on methodology. Self-reported was ~40%; IPEDS (single-category) was lower. The C/O 2028 IPEDS figure of 33% Asian appears low due to the shift to single-category counting. The C/O 2029 self-reported figure of 43.9% represents the clearest increase.

Stanford saw a sharp Black enrollment drop in Year 1 (9% to 4-5%) that persisted in Year 2 (5.8%). Hispanic enrollment continued declining in Year 2 (15% to 12.4%). The C/O 2029 has the largest Asian and White percentages in three years.

Sources: Stanford Daily (Nov 2025), Stanford Report, Stanford Facts.

MIT

*MIT switched to IPEDS methodology (single-category counting) for C/O 2029, which reduced the Asian percentage from 47% back to 38% by reclassifying multiracial Asian students. Under comparable methodology, the actual Asian increase is likely 25-35% from baseline.

MIT saw the most dramatic Black enrollment decline: 15% to 5% in Year 1, barely recovering to 6% in Year 2. The 60% drop is the largest among HYPSM. MIT's unique position as a STEM-focused institution with no legacy preference and test-required policy may explain the outsized effect. Pell-eligible students increased from 24% to 27%, suggesting socioeconomic diversity efforts partially compensate.

Sources: The Tech (Nov 2025), MIT Admissions class profile.

HYPSM Summary

Tier 2: Ivy+ (Columbia, UPenn, Brown, Dartmouth, Cornell, Duke, UChicago, Caltech)

Columbia

Columbia saw the largest absolute Black enrollment drop in the Ivy League: 20% to 12% (-40%). Asian enrollment surged from 30% to 39% (+30%). Columbia has not released C/O 2029 demographics.

Sources: Columbia Spectator (Sep 2024).

Brown

*C/O 2029 CDS data not yet released; poll data from Brown Daily Herald suggests partial recovery. CDS and poll data diverge because polls allow multiple-race identification while CDS uses single-category IPEDS.

Brown showed one of the stronger recovery signals in Year 2, with poll data suggesting Black enrollment rebounding from 8% to 12%. If confirmed by official CDS data, this would be notable. The discrepancy between CDS (single-race) and poll (multi-race) data is large: C/O 2027 Black was 9% CDS vs 13% poll.

Sources: Brown Daily Herald (Sep 2025), Higher Ed Dive.

Cornell

Cornell had the most dramatic Black enrollment decline in the Ivy League: 11.7% to 4.3% (-63%) in Year 1, with minimal recovery to 4.8% in Year 2. Hispanic enrollment also dropped sharply (-34% from baseline) though showed partial Year 2 recovery. Notably, Asian enrollment slightly decreased, possibly due to larger class size and international student composition.

Sources: Cornell Daily Sun (Nov 2025), Cornell Chronicle.

Duke

Duke switched to IPEDS methodology for C/O 2029, making comparison with prior years unreliable. The 7% Black figure is notable but cannot be directly compared to pre-SFFA baselines without consistent methodology. Duke delayed releasing demographics by 2 months, publishing only after federal reporting pressure.

Sources: Duke Chronicle (Dec 2025), Duke Today.

Dartmouth

Dartmouth declined to release its full Class of 2029 demographic profile. The only data released: 23% U.S. underrepresented students of color. Without historical comparison data under the same methodology, no meaningful Year 2 analysis is possible.

Sources: The Dartmouth (Oct 2025).

UChicago, Caltech

Neither UChicago nor Caltech has released detailed racial demographic breakdowns for C/O 2029.

Tier 2 Summary

Where data is available, Tier 2 effects are as large or larger than Tier 1:

Tier 3: Near-Ivy (JHU, Northwestern, Vanderbilt, Rice, Notre Dame, Georgetown, CMU, WashU)

Vanderbilt

Vanderbilt had one of the largest Black enrollment drops (-47% in Year 1) with no recovery in Year 2 (5.9%). The school added a supplemental identity essay for C/O 2029 applications, but this did not reverse the trend. Overall first-year class racial diversity dropped 21% in Year 1.

Sources: Vanderbilt Hustler (Nov 2024, Oct 2025), Inside Higher Ed.

Johns Hopkins

JHU admitted "significantly fewer students from underrepresented groups in fall 2024." The 4% Black enrollment is notably low. Pre-SFFA baseline unavailable from the same source, but historically JHU had ~8-10% Black enrollment.

Sources: JHU Hub (Dec 2025).

Georgetown

Georgetown has not released detailed C/O 2029 demographics. For C/O 2028, 26% Asian and 9% Black among self-reporting students was reported.

Northwestern, Rice, WashU, Notre Dame, CMU

Limited or no demographic data released for C/O 2029. Northwestern released demographics in January 2026 (42% White, 26% Asian, 17.5% Hispanic, 15.4% Black for overall undergrad body, not incoming class specifically). WashU noted "varied fluctuations" without specifics.

Tier 4: Selective Privates

Amherst College

*Estimated from prior CDS data. Amherst's Year 2 shows a strong recovery for Black (3% to 6%) and Hispanic (8% to 16%) under IPEDS, with Asian dropping from 20% to 12%. Self-reported data shows even larger recovery: Black 12%, Asian 20%.

Sources: The Amherst Student (2025).

Williams College

Williams was an outlier: Black and Hispanic enrollment slightly increased in Year 1, while Asian and White enrollment decreased. This is the opposite of peer trends and may reflect Williams' strong financial aid program (all-grant, no loans) attracting more diverse applicants. C/O 2029 data not yet released.

Sources: The Williams Record.

Middlebury College

Middlebury saw a 9 percentage point drop in students of color (35% to 26%, a -26% decline), one of the sharpest among LACs.

Sources: The Middlebury Campus (Sep 2024).

NYU

NYU saw large drops: Black -43%, Hispanic -33%. C/O 2029 data not released.

Sources: Washington Square News (Oct 2024).

Tufts

Black enrollment dropped 36%. Limited additional data.

Sources: The Tufts Daily.

Emory

Emory saw a smaller decline (-12%) than peers, possibly due to its location in Atlanta and strong Black student community.

Sources: Inside Higher Ed.

Pomona College

No disaggregated racial data released.

Sources: Pomona College news.

Tier 5-6: Public Universities

System-Wide Trends

The EdWorkingPapers study (ai26-1392) covering 3,200+ institutions and 3M+ freshmen found:

• Highly selective (all): URM freshmen down 7%; Black down 16.3%; Hispanic down 1.8%
• Ivy Plus specifically: URM down 18.9%
• Public flagships: URM enrollment UP 8% (vs. 3.2% overall growth)
• Public 4-year (non-selective): Black/Hispanic up 5.9%
• Private 4-year (non-selective): Black/Hispanic up 5.8%

The Cascade Effect

High-achieving URM students redistributed downward in selectivity: - SAT 1550-1600 URM students: shifted from Ivy Plus to other highly selective and selective publics - SAT 1400-1550 URM students: shifted out of Ivy Plus into selective publics - Net effect: diversity was redistributed across the system, not eliminated - At universities with >80% graduation rates: Black enrollment down 1.6pp, Hispanic down 1.0pp

Specific Public Universities

UVA (C/O 2028): - Hispanic: 7.2% to 9% (+25%) - Black: 8.2% to 7.2% (-12%) - Asian: 21.6% to 20% (-7%)

UC System (2024-25): - UCLA and Berkeley both saw small increases in Black and Hispanic enrollment - In-state enrollment rose to ~80% - UC system is test-blind, limiting comparability

Notable flagship examples (from IPEDS data): - LSU: Black first-year enrollment up 30% - U. Mississippi: Black enrollment up 50% - UT Knoxville: Hispanic enrollment up 33%+ - U. South Carolina: Hispanic enrollment up 33%+

Public University Assessment

The sim's T5/T6 URM multipliers of 1.30/1.25 are too aggressive. The observed +8% URM increase at flagships translates to a multiplier of approximately 1.08. The very large increases (LSU +30%, Ole Miss +50%) are at less selective publics not in the simulation. For the simulation's selective public colleges (UVA, UCLA, Michigan, etc.), effects are modest.

Cross-Cutting Findings

Application Behavior Changes

The demographic shifts are not solely driven by admission probability changes. Several confounding factors affect enrollment:

1. Application steering: URM students are less likely to apply to schools perceived as hostile post-SFFA
2. Yield changes: Admitted URM students may choose different schools when the campus feels less diverse
3. Self-reporting decline: Rising "unknown race" percentages hide true demographic composition
4. Test-required policies: Schools returning to test-required (Harvard, MIT, Yale) saw 10-15% application drops, disproportionately from populations with lower test access

Year 2 vs Year 1 Patterns

Three distinct patterns emerged in Year 2:

1. Continued decline: Harvard Black (14% to 11.5%), Stanford Hispanic (15% to 12.4%), Yale Hispanic (19% to 13%)
2. Delayed cliff: Princeton Black (9% to 5%), showing no Year 1 effect but dramatic Year 2 drop
3. Partial recovery: Cornell BHI (15.7% to 18%), Amherst Black (3% to 6%), Brown Black (polling suggests rebound)

Differentiated URM Effects

The sim currently treats all URM groups with a single multiplier. The data shows Hispanic and Black students are affected differently:

This suggests the simulation should potentially split URM into separate Black and Hispanic multipliers, as the effects differ by a factor of ~2x.

Data Availability Summary

Upcoming Data Sources

1. IPEDS ACTS (Admissions & Consumer Transparency Supplement): Data collection opened December 2025, deadline March 18, 2026. First-ever requirement for all selective institutions to report applicant data disaggregated by race-sex pairing with 6 years of historical data (2019-2025). This will be transformative. Expected public release: Fall 2026.

2. CDS 2024-25: Several schools have released (Cornell, Harvard, Stanford, CMU, Columbia). Others forthcoming through spring 2026.

3. EdWorkingPapers cascade study: Full paper with institution-level data pending publication.

Sources

• James Murphy Post-SFFA Enrollment Tracker (2025 Update)
• Inside Higher Ed: Post-SFFA Minority Enrollment Increased at Flagships
• Harvard Crimson: Class of 2029 Admissions Data
• Harvard Magazine: Class of 2029 Reflects Shifts
• Yale Daily News: Small racial shifts in second class
• The Daily Princetonian: Five percent of Class of 2029 identify as Black
• Stanford Daily: Class of 2029 enrollment data
• The Tech (MIT): Class of 2029 demographics
• Cornell Daily Sun: Hispanic, Black enrollment increases Year 2
• Columbia Spectator: Black enrollment drops
• Brown Daily Herald: Black, Hispanic students rebound
• Vanderbilt Hustler: Racial diversity drops 21%
• Duke Chronicle: Class of 2029 profile
• JHU Hub: Class of 2029 demographics
• Amherst Student: First-year class data
• Williams Record: Demographics shift slightly
• Middlebury Campus: Diversity drop
• Washington Square News (NYU): Black and Latino drops
• Inside Higher Ed: Fewer colleges publicized demographic data
• EdWorkingPapers: College Enrollment Patterns After SFFA v. Harvard
• Georgetown Feed: Cascade Effect

Some sections containing simulation-specific implementation details have been omitted from this public version. The research data and analysis above is based on publicly available sources.