And Barak Obama was a transfer into Columbia - not part of their freshman class.

If you spent time at an elite campus, you'd know that they do a pretty great job of picking a school full of very smart, conscientious students. The problem is that those classes don't look like America - not nearly enough students from working class families of all colors, despite very generous financial aid. Using test scores doesn't do any better because those are even more skewed towards the upper middle class.

There is plenty of institutional research on how well students do - they pretty much all graduate and they do really well in grad school and in their careers. But, there is no way to predict a black swan like Steve Jobs (Reed) or Barack Obama (Columbia/Harvard) or Sonia Sotomayer (Princeton/Yale) over an ordinary alum like me.

No one is going to mess with success with a random experiment. Would Apple or McKinsey or Goldman randomly hire people?

Anonymous wrote:

Anonymous wrote:It sounds reasonable. What don't you like about it?

OP here. It just seems weird and arbitrary. If you read Moneyball by Michael Lewis it is kind of obvious that with this much applicant data it should be possible to design a system that predicts success better than telling admissions people to go out and pick a couple of thousand kids (out of 10,000 with similar stats?) who seem interesting to them.

Is there any follow up? What breaks the monotony for someone reading 2,000 applications from kids probably starts being very bizarre after a while. Do they track whether their hunches about the kids turned out to be correct? Do they ever do experiments --- admit 200 kids who write how much they love chemistry and 200 who write about about phantom farting, or whatever, perhaps just a control group chosen randomly from a SAT/GPA band, and then see which group turned out better? Is it all basically seat of the pants?

That's why people say top-tier college admission is a crapshoot!

Anonymous wrote:It sounds reasonable. What don't you like about it?

OP here. It just seems weird and arbitrary. If you read Moneyball by Michael Lewis it is kind of obvious that with this much applicant data it should be possible to design a system that predicts success better than telling admissions people to go out and pick a couple of thousand kids (out of 10,000 with similar stats?) who seem interesting to them.

Is there any follow up? What breaks the monotony for someone reading 2,000 applications from kids probably starts being very bizarre after a while. Do they track whether their hunches about the kids turned out to be correct? Do they ever do experiments --- admit 200 kids who write how much they love chemistry and 200 who write about about phantom farting, or whatever, perhaps just a control group chosen randomly from a SAT/GPA band, and then see which group turned out better? Is it all basically seat of the pants?

OP -- Yup, sounds like what everyone has told me about how it works. And sounds like Tina Fey 's version too in the movie Admission (see it, if you haven't yet).

Was this the essay where he talks about how they admitted the girl who wrote about farting? makes you wonder why you wasted all that money on music lessons, etc. if that's what it takes to get in.

Depends on the school, but, yes, this sounds realistic. There are some variations (e.g. UChicago gives more weight than most to essays), but the process is such that standing out really matters and the challenge is finding an effective/attractive way to do so. Stats establish that you're qualified (even well-qualified), but they don't get you admitted. And what *does* get you in varies from school to school and from pool to pool.

When so many of the metrics are equal - near perfect SATs and GPAs - this other stuff does indeed come to the forefront.

MIT and Carnegie Mellon are very similar. CalTech just runs the numbers and ranks the applicants. The state schools use numbers but are a couple standard deviations lower than the others.

It sounds reasonable. What don't you like about it?

Yes it does. Public schools are more stats driven.

Came across this:

http://nypost.com/2016/02/07/former-yale-admissions-officer-reveals-secrets-of-who-gets-in/

Does anyone know whether this is just whimsy or whether it describes how the process actually works at these schools? Are the top tech or state schools more metrics driven?