This series takes a school's published acceptance rate apart using the school's own Common Data Set filing — splitting the headline number into in-state, out-of-state and international pools, or into early and regular rounds, so a family can see which rate actually applies to them. For most schools we cover, the filing supports that. For nine of the schools families search for most, we opened the current filings in August 2026, and at seven of them the residency table is submitted with no numbers in it.
So when you go looking for the in-state or out-of-state breakdown behind Harvard's or Yale's acceptance rate and cannot find it, you are not looking in the wrong place. The number does not exist publicly. The school did not report it.
The Residency Table Is Optional, and That Matters
Section C1 of the Common Data Set asks every college for applicants, admits and enrolled students for the entering class. Underneath that block sits a supplementary table breaking the same three counts into In-State, Out-of-State, International and Unknown columns.
That supplement is optional. In all nine filings we opened, it is introduced by the same printed sentence: "If available, please provide residency breakdowns for total applicants, admits, and enrolled students." A school that leaves it empty has answered the form correctly. No rule is broken and no standard is missed, and we are not going to speculate about why any school leaves it blank, because no school publishes a reason.
What a blank table does mean is narrower: no residency breakdown for that school exists in the public record.
Turning research into a plan?
Get a personalized roadmap from a near-peer mentor. Free 25-minute call, no commitment.
What the Nine Filings Show
These are nine specific schools, picked because they are the highest-search-volume institutions in our project, not a random sample and not a survey of American higher education. Every figure below comes from the school's own current Common Data Set, read in August 2026.
School | Residency table in Section C1 | What the filing does report |
|---|---|---|
Harvard | Detail columns empty | Total only: 47,893 applied, 2,003 admitted, 1,675 enrolled |
Stanford | Detail columns empty | Total only: 60,646 applied, 2,302 admitted, 1,839 enrolled |
Yale | Entirely empty, including Total | 22,819 men and 27,445 women applied |
MIT | Entirely empty, including Total | Total in C1: 29,281 applied, 1,334 admitted |
UChicago | Entirely empty | Sex rows sum to 45,533 applied, 2,039 admitted |
Caltech | Detail columns empty; Total reads 0 | C1 sex rows sum to 11,285 applied, 428 admitted |
Carnegie Mellon | No residency supplement in the filing | 34,867 applied, 3,859 admitted |
Georgetown | Complete | 243 in-state, 22,653 out-of-state, 3,926 international |
Princeton | Complete | 4,602 in-state, 27,134 out-of-state, 10,567 international |
Harvard's filing is the cleanest illustration, because there is nothing wrong with it. Its In-State, Out-of-State, International and Unknown columns are empty and its Total column is filled: 47,893 applied, 2,003 admitted, 1,675 enrolled. Those totals reconcile exactly against the sex rows above them — 24,000 men plus 23,872 women plus 21 students of unknown sex is 47,893, and 968 plus 1,035 is 2,003. The filing is consistent to the applicant. It simply reports nothing by residency, so 4.18% overall is the only admit rate it supports. Stanford's behaves and reconciles the same way, with 31,609 plus 29,011 plus 26 giving its 60,646 total.
Yale's is the sharpest contrast, because of what sits a few lines below the blank table. Yale's residency block has nothing in it at all, not even a Total. In the same filing, the waiting-list block is completed to the person: 944 applicants offered a place, 738 accepting a place, 32 admitted from the list. And in Section C21, where the form asks for early decision applications, the field still holds the Word template's own unfilled prompt — the words printed in the filing are "Click or tap here to enter text." Yale answers "No" to offering an early decision plan, so there is no count to enter there and nothing is missing. What the placeholder shows is an untouched form field in a document where other fields were filled with care.
Caltech's is the one filing where the blank table produces something that cannot be read at face value. Its four residency columns are empty and its Total column reads 0 for applicants, admits and enrolled — while Section C1 of the same filing reports 7,772 men and 3,513 women applying, and 224 and 204 admitted. We are reporting what the filing shows, not what was intended by it.
Two of the Nine Publish the Whole Breakdown
This is not a claim that highly selective schools uniformly leave this blank. Two of the nine fill the table in completely, and both reconcile exactly against their own Section C1 totals.
Princeton's Fall 2025 table reports 4,602 in-state applicants, 27,134 out-of-state and 10,567 international, summing to 42,303 — exactly what its sex rows sum to. Its admits split 229, 1,398 and 241, summing to 1,868. That supports a real decomposition: 4.98% for New Jersey residents, 5.15% out-of-state and 2.28% international, against 4.42% overall.
Georgetown's is equally complete: 243 in-state applicants with 48 admitted, 22,653 out-of-state with 3,244 admitted, and 3,926 international with 326 admitted, summing to the 26,822 and 3,618 its Section C1 reports. That works out to 19.75% for District of Columbia residents, 14.32% out-of-state and 8.30% international, against 13.49% overall — though the in-state figure rests on a pool of only 243 applicants and moves a great deal on small changes.
The counterweight beyond these nine is larger still. Public flagships and mid-tier private universities fill this table in routinely, which is the reason our Real Acceptance Rates project exists at all. The gap here is between individual filings, not between elite schools and everyone else.
Early Rounds Are a Different Problem
If you go looking for an early-round breakdown at these schools, you hit something else entirely, and it is structural rather than institutional.
The Common Data Set asks for early decision counts in Section C21. It asks for nothing comparable about early action — Section C22 asks only whether the school has an early action plan, its closing and notification dates, and whether it is restrictive. There are no count fields. Seven of these nine run early action rather than early decision, so their C21 correctly reads "No" and no early-round rate is computable from their filings. That is the form's design, not a choice any of them made.
Only two of the nine have an early decision plan at all, and they diverge. Carnegie Mellon answers "Yes" and fills the counts in: 2,680 early decision applications for the Fall 2025 entering class and 553 admitted, a 20.63% early decision rate against 11.07% overall. UChicago answers "Yes" and leaves the closing date, the notification date and both count fields empty.
One caution on any early decision figure, Carnegie Mellon's included: Section C21 is the section most prone to carrying a prior year's block forward, which we documented across 23 schools in our investigation of the Common Data Set's Early Decision problem. Check the cohort label printed on the block itself before trusting it.
MIT publishes the most of the nine, and none of it comes from its Common Data Set. Its residency table is blank in both its 2025-26 and 2024-25 filings. But on its own admissions statistics page MIT publishes a full split for the Class of 2029: 12,052 early action applicants with 721 admitted, and 17,229 regular action applicants with 603 admitted including deferred students. That gives a 5.98% early action rate against 3.50% regular, and the parts close on the whole — 12,052 plus 17,229 is exactly the 29,281 in its Common Data Set, and 721 plus 603 plus its 10 wait-list admits is exactly 1,334. MIT's page says outright that these statistics go beyond what it contributes to the Common Data Set. A school can publish more than the form asks.
Why a Number You Find Elsewhere Isn't a Substitute
If a ranking site, a consultant or an admissions blog shows you an in-state or early-round figure for one of the seven schools above, it did not come from that school's Common Data Set, because the Common Data Set does not contain it. It came from something else — a modelled estimate, a different survey, a figure carried forward from an older cycle, or a number attributed to an unnamed source.
That last category is worth naming. A figure quoted in a single article and attributed to an anonymous spokesperson cannot be checked by you or by us against anything the school itself publishes. It may well be accurate. It is not verifiable, and a number you cannot verify should not anchor a decision about where to apply.
The Check You Can Run in Five Minutes
- Search the school's name plus "Common Data Set" and open the current filing on its institutional research office's own site. Confirm the year on the cover.
- Go to Section C1 and find the residency supplement below the main table. If the In-State, Out-of-State and International columns are empty, no residency breakdown exists for that school, and any residency figure you have seen quoted for it came from somewhere other than this filing.
- Check that the detail rows which are filled actually sum to the totals. At Harvard, Stanford, Princeton and Georgetown they do, exactly. That is a fast test of whether a filing hangs together.
- Go to Section C21. If the school answers "No" to early decision, no early-round rate is computable from the filing, because Section C22 collects no early action counts. If it answers "Yes" and the count fields are empty, the same is true.
- Before trusting an early decision figure that is filled in, read the cohort year printed on the C21 block itself, not the filing's cover year, and compare it against the prior year's filing.
Where to Go From Here
Not knowing a school's residency split is survivable. Believing an invented one is not. Where a breakdown does not exist, build around the rates that do and around the parts of your application you control.
Our free college list builder is built on rates we have pulled and checked filing by filing, and our application deadline tracker keeps early action, early decision and regular deadlines straight across a list that mixes all three, which most lists containing these nine schools will. If you want to work through how much weight a published rate deserves at the schools you are applying to, a free strategy call is a good place to start.
Want guidance like this from a mentor like Boris?
Our near-peer mentors have walked this exact path. Free 25-minute call, no commitment.
Frequently Asked Questions
Why can't I find Harvard's in-state and out-of-state acceptance rates?
Because Harvard does not report them. The residency breakdown in the Common Data Set is an optional supplement, introduced in the form itself with the words "If available, please provide." Leaving it blank is a permitted answer that breaks no rule or standard, and we are not suggesting otherwise. Harvard's 2025-2026 filing reports 47,893 applicants and 2,003 admits in total, with no residency detail beneath them.
Where do the in-state acceptance rates I have seen quoted for these schools come from?
Not from their Common Data Sets, which do not contain them. Any residency figure presented for one of the seven schools here was derived from something else — a model, a different survey, an older cycle, or an unnamed source. Ask which before you rely on it.
Do any highly selective schools publish the full breakdown?
Yes. Princeton and Georgetown, two of the nine here, fill the residency table in completely, and both reconcile exactly against their own Section C1 totals. Princeton's Fall 2025 split runs 4.98% in-state, 5.15% out-of-state and 2.28% international against 4.42% overall; Georgetown's runs 19.75%, 14.32% and 8.30% against 13.49% overall. Plenty of public flagships fill it in too, which is why the rest of this series is possible.
Why can't I find an early action acceptance rate in a Common Data Set?
Because the form does not ask for one. Section C21 collects early decision counts, but Section C22, covering early action, asks only whether the plan exists, its dates and whether it is restrictive — there are no count fields. Seven of the nine schools here run early action rather than early decision, so nothing about their early round appears in their filings. The exceptions are Carnegie Mellon, which reports 2,680 early decision applications and 553 admits for Fall 2025, and MIT, which publishes its early action split on its admissions statistics page rather than in its filing. Sources: Common Data Set 2025-2026 filings, Sections C1, C2, C21 and C22, published by Harvard, Stanford, Yale, MIT, the University of Chicago, Caltech, Carnegie Mellon, Georgetown and Princeton, each opened and read in August 2026; MIT's 2024-2025 Common Data Set; MIT Admissions' published admissions statistics for the Class of 2029.
