Selective colleges insist they read every applicant in context. The context that matters most, it turns out, is often the one printed at the top of the envelope: a state, a ZIP code, a high school. The map is not the territory — but in admissions, it might as well be.
Admission deans speak about geographic diversity the way Bordeaux producers speak about terroir — as if it were a delicate balance carefully tended. The data tells a different story.
Across the most selective private colleges, five states supply the bulk of every entering class. New York, California, Massachusetts, New Jersey, and Connecticut together represent roughly a quarter of the US population — but they routinely furnish 40 to 50 percent of the seats at HYPSM and the Ivy League.
Connecticut is the cleanest example: 1.1 percent of Americans, 2 to 4 percent of every elite class. That is not diversity; that is gravity.
If state-level skew is the symptom, regional pull is the mechanism.
Plot the regional makeup of any selective college's class and a pattern emerges: the largest single bar is almost always the region the college sits in.
Harvard's class is 39 percent Northeastern. Stanford's is 40 percent Western. Duke draws 28 percent from the South. Notre Dame and Northwestern, the two big Midwestern privates, each pull more than a quarter from the Midwest — far above the 10 to 14 percent typical at coastal peers.
The smallest figures are nearly always the same: 10 to 12 percent from the Midwest at the Northeastern Ivies, despite the region holding 20 percent of the US population.
Northeasterners go to Northeastern colleges. So who wins by being far away?
Every elite school wants to write “students from all 50 states” on its class profile. That sentence is a quota, however softly stated. And in low-population states, it bends the math.
In 2015, Brown received 23 applications from Montana. It accepted seven. That is a 30 percent acceptance rate for a school whose overall rate that year was 8.5 percent — a 3.5× relative advantage that admissions consultants still cite.
The boost is not unconditional — applicants still have to be competitive — but the math is unforgiving: an Ivy receives 5,000 applications from New York and perhaps 30 from Wyoming. The marginal Wyoming applicant is far more likely to be needed.
Public universities have a different geography problem — one written into law.
At public flagships, geography is policy. The North Carolina Board of Governors fixed the in-state floor at 82 percent in 1986 and has policed it with budget penalties ever since. UT Austin's law sets the cap at 90 percent. Virginia's legislature mandates two-thirds in-state at UVA.
The University of California's Board of Regents wrote a system-wide nonresident cap of 18 percent; Berkeley and UCLA were grandfathered above it but are trending back to 80 percent in-state by 2025. Wisconsin removed its 25 percent out-of-state cap in 2015 and watched OOS share double.
Where there is no formal cap — Michigan, Purdue, UIUC — the in-state share still tends to land in the 45 to 55 percent range, mostly for political and budgetary reasons.
A mandate is one thing. The acceptance gap it produces is another.
Compare the in-state and out-of-state acceptance rates at a public flagship and you can read the strength of the geographic preference directly.
At UNC, Wisconsin, UF, and UT Austin, in-state applicants are accepted at 3.5 to 4.75 times the out-of-state rate. Georgia Tech runs a Georgia-only Early Action round that admits at 33 percent; the non-Georgia EA round admits 8 percent.
Michigan and UIUC sit in the middle. UC Berkeley's gap is the tightest at 1.4× — the OOS pool there is so self-selected and academically extreme that the rate gap nearly closes. Virginia Tech is the lone inversion: OOS acceptance (59%) actually exceeds IS (48%).
If the public-school question is in-state vs. out, the private-school question is rural vs. urban.
Roughly 19 percent of Americans live in rural communities. They send a much smaller share of children to the most selective colleges — 9 percent at Princeton, 5 to 6 percent at Yale.
The barriers are familiar: rural schools average one counselor per 500-plus students versus 1:30 at coastal preps, fewer AP courses, no calculus at some, and a thin information pipeline about elite-college aid. Even when admitted, rural students yield at lower rates.
Two schools stand out the other way. Dartmouth and Duke each draw 15 percent of their classes from rural America, and both are members of the STARS network — a coalition of 33 elite colleges that together extended 11,000+ admissions offers to rural and small-town students for the Class of 2028 alone.
The domestic geography is competitive enough. The international pool plays by different rules entirely.
Roughly 1.18 million international students studied in the US in 2024-25, an all-time high. India sent 363,019, up 10 percent year over year. China sent 265,919, down 4 percent.
The international share at selective colleges is a window onto how much of the cohort is competing internationally for a small slice of seats. NYU's class is roughly a quarter international. USC's is 26 percent. Carnegie Mellon's is 22 percent. Many LACs run 10 to 16 percent.
Within the international pool, geography compounds again. A student from China competes against thousands of other Chinese applicants for a handful of slots; a student from Bhutan or Moldova faces almost no competition from compatriots. The Montana effect, in other words, has an international cousin.
All of it — state skew, regional gravity, rural underrepresentation — flows from one place: high schools.
Christine Glasener's 2022 University of Michigan dissertation mapped 3,200 secondary schools to 76 selective colleges. Less than 20 percent of the country's roughly 19,000 high schools qualified as “feeders.” The most connected high school had ties to 58 of the 76 colleges studied.
These feeders cluster in a handful of metros: New York, Boston, Bay Area, Washington D.C., Los Angeles, and the New England boarding-school belt of Massachusetts, Connecticut, and New Hampshire. Andover sends roughly 221 students to the Ivy League each year; Choate, 212; Lawrenceville, 179.
Approximately half of Harvard's class comes from just ~200 high schools out of more than 27,000 in the United States. The Big Five states are not an accident. They are the home addresses of the feeder schools.
Selective admissions, viewed from above, is not really a meritocracy of individuals. It is a weighted lottery of high schools, with state caps layered on top and a handful of geographic boosts thrown in to keep the “all 50 states” line on the brochure honest.
None of this is hidden. UNC's 82 percent floor is in policy 700.1.3. UC Berkeley's nonresident cap is on the Regents' website. Brown's regional officers are listed by territory. The Glasener dissertation is publicly archived. The system simply doesn't advertise that your address can move your odds further than your essay can.
The College Monte Carlo simulation models these geographies explicitly: an in-state logit boost ranging from +0.34 at Berkeley to +1.56 at UNC, an underrepresented-state hook for Montana, Wyoming, and the Dakotas, and feeder-school multipliers for the 200 schools that supply half the Ivy League. The point is not that geography determines admission. The point is that it is doing more work than the marketing suggests.