There is also a problem in that the right tail of the academic distribution has been truncated by high school grade inflation and the recentering of the SAT. There is a wide range of ability within the top 1.0% and it makes a difference. Harvard's basketball coaches are not impressed by high school players in the top 1.0% (roughly top 5,000 nationally, plus many non-U.S. players). Their recruiting efforts are directed at the top 0.1%. It is now much more difficult now for admissions committees to tell the difference between those applicants who are strong academically and those who are exceptional.
40 years ago Robert Klitgaard wrote a thoughtful and rigorous book called Choosing Elites. At the time Klitgaard was a professor and Admissions Chairman at the Kennedy School at Harvard, and also Special Assistant to Harvard's President, Derek Bok. Based on a careful analysis, he reached a conclusion similar to Prof. Deming's: "We have some confidence that those who do better academically will also do better in later life. But beyond this, current nonacademic information cannot be shown to help much in selecting among applicants at the right tail those through whom a top university will create the greatest social value." He also wrote ""Intuitively, some readers may think they can learn a lot about candidates from essays , but studies do not support the predictive power of such measures, and again one worries about dissimulation."
Every time I delve into this paper (which, by the way, has managed to be one of the most captivating reads for me in recent times - a significant feat given the current status of things), I find my belief in Ivy-plus institutions as vehicles for socioeconomic mobility waning.
Two marginal improvements seem self-evident:
- Abolishing the holistic admission process in favor of returning to a purely test-based selection, and
- Doubling the number of admission spots, which should be well within the means of the Ivy-plus endowments.
Sadly, the will to implement such changes seems lacking.
Yes. Also, I must once again point out that there's an inherent restriction of range issue with the SAT - kids who do poorly on the SAT and would go on to do poorly in college GPA are kids who don't go to college and so aren't in the sample. When you correct for restriction of range the SAT-GPA correlation is consistently robust. This has been a known issue for decades but nobody talks about it, mostly because they don't want to acknowledge the strength of the test.
Hi Freddie, thanks for sending the interesting article and for reading. You might be interested in some appendix figures that are buried in the back of the paper (Figures A.25 and A.26), which show supporting evidence for your claim that the SAT is an important predictor. We don't look at college GPA but rather future earnings, grad school attendance, and other long-run outcomes.
It seems difficult to escape the conclusion already drawn by Pier that the situation could be unambiguously improved by:
- basing admission only on test scores / exams;
- doubling the number of places at Ivy-plus (Brad De Long compared the Ivy-plus to the most profitable firms under the self-management system in communist Yugoslavia, which would shun opportunities to expand because the existing staff would not see the benefit of it).
I like very much where you finish here, which is that the ambiguity of "holistic admissions" is often seen by a salve inside academic institutions ("nobody has to feel bad, because nobody knows exactly why we said no") and instead it is actually intensely corrosive to applicants regardless of their qualifications, their socioeconomic class, their credentials (of any kind); it's contributing to why elite institutions are more and more disliked over time. We're making decisions that we believe are important and material to the future of ambitious young people and their families whose basis is completely mysterious to the applicants--and the applicants often are able to see by comparison how perverse and arbitrary the decisions seem to be when they look at where they were accepted and where they weren't. Transfer students often get an especially sharp look at it because they sometimes move to an institution that said, by rejection, "You aren't going to work out here" only to find the truth is "You actually excelled here".
The sum total impact of that is a really serious loss of confidence that is hurting all those institutions.
"About 2 million high school seniors take the SAT each year (a bit less in recent years). Another 1.5 million take the ACT. By definition, 35 thousand of them score in the top 1 percent"
The pools are overlapping, some kids take both tests, and those who do well on one likely do well on the other. So the top 1% of test takers is a smaller pool than 35,000.
But, I think you should be looking at the entire cohort of 18 year olds, which is about 4 million. Assuming that intelligence is the relevant trait which is roughly measured by test results, and assuming that it is randomly distributed across the population the relevant universe is 40,000 kids. Your 35,000 is close but I think your reasoning is off.
The other point is that a lot of seats in your top schools are not taken by the smartest kids. There are athletes, legacies, children of prominent politicians, faculty brats who are smart enough to do the work, but fall outside the top 1% . My guess is that the magic circle colleges are really only admitting less than 10,000 of the top 1%.
This piece nicely distinguishes the elite admissions opportunities created by the combination of wealth and geography. But unremarked upon is the failure of “holistic” to emphasize admission of talented kids from poor high schools. Now that high school rankings exist for every high school in the US, one could analyze how well-represented are bottom quartile high schools in Ivy-plus student populations.
There is also a problem in that the right tail of the academic distribution has been truncated by high school grade inflation and the recentering of the SAT. There is a wide range of ability within the top 1.0% and it makes a difference. Harvard's basketball coaches are not impressed by high school players in the top 1.0% (roughly top 5,000 nationally, plus many non-U.S. players). Their recruiting efforts are directed at the top 0.1%. It is now much more difficult now for admissions committees to tell the difference between those applicants who are strong academically and those who are exceptional.
40 years ago Robert Klitgaard wrote a thoughtful and rigorous book called Choosing Elites. At the time Klitgaard was a professor and Admissions Chairman at the Kennedy School at Harvard, and also Special Assistant to Harvard's President, Derek Bok. Based on a careful analysis, he reached a conclusion similar to Prof. Deming's: "We have some confidence that those who do better academically will also do better in later life. But beyond this, current nonacademic information cannot be shown to help much in selecting among applicants at the right tail those through whom a top university will create the greatest social value." He also wrote ""Intuitively, some readers may think they can learn a lot about candidates from essays , but studies do not support the predictive power of such measures, and again one worries about dissimulation."
Abolish admissions offices and replace them with a lottery.
Stay tuned!
Every time I delve into this paper (which, by the way, has managed to be one of the most captivating reads for me in recent times - a significant feat given the current status of things), I find my belief in Ivy-plus institutions as vehicles for socioeconomic mobility waning.
Two marginal improvements seem self-evident:
- Abolishing the holistic admission process in favor of returning to a purely test-based selection, and
- Doubling the number of admission spots, which should be well within the means of the Ivy-plus endowments.
Sadly, the will to implement such changes seems lacking.
Yes. Also, I must once again point out that there's an inherent restriction of range issue with the SAT - kids who do poorly on the SAT and would go on to do poorly in college GPA are kids who don't go to college and so aren't in the sample. When you correct for restriction of range the SAT-GPA correlation is consistently robust. This has been a known issue for decades but nobody talks about it, mostly because they don't want to acknowledge the strength of the test.
https://freddiedeboer.substack.com/p/restriction-of-range-what-it-is-and-why-it-matters
Hi Freddie, thanks for sending the interesting article and for reading. You might be interested in some appendix figures that are buried in the back of the paper (Figures A.25 and A.26), which show supporting evidence for your claim that the SAT is an important predictor. We don't look at college GPA but rather future earnings, grad school attendance, and other long-run outcomes.
They also don't want to acknowledge that SAT + GPA is a better predictor than GPA alone.
It seems difficult to escape the conclusion already drawn by Pier that the situation could be unambiguously improved by:
- basing admission only on test scores / exams;
- doubling the number of places at Ivy-plus (Brad De Long compared the Ivy-plus to the most profitable firms under the self-management system in communist Yugoslavia, which would shun opportunities to expand because the existing staff would not see the benefit of it).
I like very much where you finish here, which is that the ambiguity of "holistic admissions" is often seen by a salve inside academic institutions ("nobody has to feel bad, because nobody knows exactly why we said no") and instead it is actually intensely corrosive to applicants regardless of their qualifications, their socioeconomic class, their credentials (of any kind); it's contributing to why elite institutions are more and more disliked over time. We're making decisions that we believe are important and material to the future of ambitious young people and their families whose basis is completely mysterious to the applicants--and the applicants often are able to see by comparison how perverse and arbitrary the decisions seem to be when they look at where they were accepted and where they weren't. Transfer students often get an especially sharp look at it because they sometimes move to an institution that said, by rejection, "You aren't going to work out here" only to find the truth is "You actually excelled here".
The sum total impact of that is a really serious loss of confidence that is hurting all those institutions.
"About 2 million high school seniors take the SAT each year (a bit less in recent years). Another 1.5 million take the ACT. By definition, 35 thousand of them score in the top 1 percent"
The pools are overlapping, some kids take both tests, and those who do well on one likely do well on the other. So the top 1% of test takers is a smaller pool than 35,000.
But, I think you should be looking at the entire cohort of 18 year olds, which is about 4 million. Assuming that intelligence is the relevant trait which is roughly measured by test results, and assuming that it is randomly distributed across the population the relevant universe is 40,000 kids. Your 35,000 is close but I think your reasoning is off.
The other point is that a lot of seats in your top schools are not taken by the smartest kids. There are athletes, legacies, children of prominent politicians, faculty brats who are smart enough to do the work, but fall outside the top 1% . My guess is that the magic circle colleges are really only admitting less than 10,000 of the top 1%.
This piece nicely distinguishes the elite admissions opportunities created by the combination of wealth and geography. But unremarked upon is the failure of “holistic” to emphasize admission of talented kids from poor high schools. Now that high school rankings exist for every high school in the US, one could analyze how well-represented are bottom quartile high schools in Ivy-plus student populations.