^^^^ Makes sense, since they’re all full of…
(I am sorry I could not resist. I’ll serve my time.)
Post- don’t apologize. That’s the natural reaction, no? 
This statement assumes the people you meet in college will be a random cross-section of the entire undergraduate student body. But it really doesn’t work like that at a large university. If you’re a student in Ross, Michigan’s business school, you’ll mostly take classes with and are much more likely to socialize with other Ross students, who as a group have academic stats almost indistinguishable from those at Penn’s Wharton School. Engineering students will naturally spend more time with other engineering students, who as a group have academic stats well above those of Michigan’s other undergraduate schools (except Ross). I got my undergrad degree at Michigan ages ago, but as a freshman I opted for the honors program (open to roughly the top 10% of the entering class), lived in honors housing, and took almost entirely honors courses my freshman year. As a sophomore I accelerated into upper-level courses, mostly with juniors and seniors, and by my third and fourth years I was taking mostly graduate-level courses where my peers were grad students in one of the most highly respected and truly elite graduate programs in my field (philosophy). And that’s certainly not unique to my major. Math whizzes take accelerated classes with other math whizzes, including elite mathematics grad students. And so on.
I don’t mean to suggest these academic units are completely walled off from each other; there’s lots of mixing in student organizations, ECs, at parties, etc. But your social circle will be far from a random cross-section of the class, and if you’re a top student in a rigorous program you’ll tend to have far more interactions with other top students than with others.
Thank you!
What @bclintonk said—and on the flip side, pretty much every school is going to have slackers (maybe, at some schools, brilliant enough slackers to still pass while coasting by, but slackers nonetheless), and in my ahem personal experience, the slackers at high-end colleges will find each other, too.
http://talk.collegeconfidential.com/parents-forum/1933632-newly-discoverd-dumb-antics-by-harvards-male-soccer-team-results-in-cancellation-of-season.html indicates that being able to get into a super-selective school does not guarantee good judgement in all matters.
The problem with all these studies is probably coming up with an adequate control group. Of those who get admitted to Harvard, roughly 80% attend. Of the remaining 20%, I don’t have the data to prove it but expect most turn it down for other ivies/MIT/Stanford/Caltech. Students who turn down Harvard for the University of Alabama are probably as rare as unicorns.
Not a scientific sample by any stretch of the imagination, but for the specific Harvard-Alabama pair, you might would find something at http://talk.collegeconfidential.com/university-alabama/1549857-what-schools-did-your-child-pass-on-in-order-to-attend-alabama-p1.html which could lead you to find how anyone who went to Alabama instead of Harvard ended up doing.
I have to say, though, if any of my kids got into both, Harvard would be nearly full pay according to their NPC—and no, we’ve got a good income, but not that good—and Alabama would be about $10k. The decision has effectively already been made—which means my kids will never be ones that turn down Harvard for someplace else, because they’re never applying to Harvard. That invisible pre-creen of applicants skews the ability to make comparisons, too.
One of the kids that was accepted to all 8 ivies last year chose a full ride with Alabama honors program instead, as discussed at http://www.businessinsider.com/ronald-nelson-turned-down-every-ivy-league-school-for-university-of-alabama-2015-5 . Of course this doesn’t happen often, as students who favor Alabama over Harvard generally aren’t going apply to spend a lot of time and money applying to Harvard as a backup in case Alabama rejects them. Instead they’ll skip the Harvard application altogether. Furthermore the schools are in different sections of the country. I’d expect the main exceptions to relate to financial surprises, such as getting a competitive Alabama scholarship or worse FA than expected at Harvard.
I expect this happens a lot. If your parents can’t or won’t pay for Harvard, even if you get admitted, there’s not much point in applying. It’s also why I think Krueger’s self revelation variable(best school applied to) is probably a proxy for having parents able to pay or at least cosign for the cost of elite colleges as much as anything else.
Harvard (and other similar “elite” colleges) isn’t exactly known for a large portion of parents cosigning loans. According to the IPEDS database, Harvard students had the 2nd lowest loan rate of all colleges in the US with a non-zero loan rate and reported ACT score. Only Hebrew Theological College was lower. Similarly I wouldn’t assume that a larger portion of families cannot pay for Harvard than other colleges. Their website claims, “Ninety percent of American families would pay the same or less to send their children to Harvard as they would a state school.” Their NPC suggests families with near the median income in the US would pay almost nothing after FA. That said, a large portion of Harvard admits come from wealthy families, so among Harvard admits, you’ll find a good portion who can find less expensive options at their state flagship, even if most US families would not.
Except the bolded does not seem to take into account that students who qualify for admission at Harvard are most likely competitive for large scholarships at lower ranked schools. Add our family to the list of won’t apply to schools like Harvard bc we cannot afford it. We would not be full pay at Harvard, but our family contribution at Harvard is still significantly beyond our ability to pay. Our kids have qualified for merit scholarships at other schools bringing our costs down to our tiny budget. Our current college student is attending on full scholarship. $0 compared to our expected contribution at schools like Harvard is a HUGE difference for our family. $0 wins. (And attending UA has been a wonderful experience for our son.)
And to tag on to what @Mom2aphysicsgeek wrote, this is actually kind of disingenuous on Harvard’s part—if you dig behind their numbers, you’ll find that they’re saying that for 90% of American families, it would cost less for their kids to go to Harvard than to the average public (that’s a big caveat—for one, we’re not just talking flagships here) without further aid.
I don’t know what the actual percentage would be, but despite the way it sounds, it isn’t true that it would be cheaper for 90% of students’ families to send their kid to Harvard than to send their kid to a public university.
Havard’s NPC lists the following net price to parents with typical assets:
$65k income – $0 cost to parents
$100k income – $5k cost to parents
For families with incomes that are less than double the US median household income like this, it can be difficult to match Harvard’s cost after FA, even with merit scholarships at other colleges. That said, yes the 90% figure includes all families in the US, not just those who are accepted to Harvard, and those accepted to Harvard often are eligible for quality merit scholarships elsewhere and often have high incomes.