💾 Archived View for dioskouroi.xyz › thread › 29410840 captured on 2021-12-03 at 14:04:38. Gemini links have been rewritten to link to archived content
-=-=-=-=-=-=-
________________________________________________________________________________
At any given university, the kids from public schools are smarter than the kids from private schools. This is a necessary consequence of private schools improving kids' admissions prospects, and parents wouldn't send their kids to private schools if they didn't.
Is there data to backup this claim?
In my experience most of the top students at top universities attended private high schools. (I went to public school)
This might be enough to make the average private school student smarter than the average public school student.
It seems likely that the median public school student is smarter than the median private school student, based on the same reasoning in the tweet (but that is less encouraging)
> This might be enough to make the average private school student smarter than the average public school student.
Note that here we're talking about something different - namely, that _at any given university_, the kids from public schools are smarter than the kids from private schools. We don't talk about all kids - only kids in universities, and there is selection filter for who gets there, so in universities the distribution of "smartness" for private or public school kids is different than in schools.
In other words, maybe you're right and the average private school student is smarter than the average public school student. But it's not the average student who gets to a university.
There was once a stable state when many universities had several candidates for each place, and could choose whom to admit. Theoretically, the better the university the bigger the pool of candidates it has. However some universities were notoriously hard to get to, so candidates didn't even bother to apply there. As a result, the candidate pools in those universities were smaller, but those candidates were of a different kind - as if they have already passed pre-filter which made them more confident that they are those special ones who has better than average chance to get to those hard to get universities.
Without any supporting evidence, I have no idea how this qualifies as news.
This Tweet is on HN because its founder, Paul Graham, wrote it. I would contend that if anyone else wrote these assertions (even if a celebrity), it wouldn't receive as much engagement.
If I understand it correctly, mostly the smartest kids from public schools get admitted to a university. On the other hand kids from private schools don't have to be that smart as their schools gave them significantly better opportunity to attend universities.
Looks like to me that the public school kids get admitted their schools despite.
If this is actually true (I don't see a citation), then it looks like a classic example of the "explaining away" effect (also know as "collider effect" or Berkson's Paradox):
https://martin-thoma.com/explaining-away/
Better explanation:
https://en.m.wikipedia.org/wiki/Berkson%27s_paradox
The premise that private school gives an admissions boost may be true but it seems to me that this argument would only necessarily be true for the marginal student, ie public school student A and private school student B both just cleared the bar for admission then student A had to rely more on their own talent. If I had to guess PG is thinking about Ivy/Stanford where admissions are so strict that almost everyone is near the margin and needs some luck to get in. My university was not competitive and the average student was comfortably over the bar so that no such relationship holds.
In some countries the most elite schools are the public high schools and universities. At those universities in such countries the kids from private schools are crowded out from the top elite universities (whose entrance is solely exam based and which are public and usually free or almost free).
So the private universities which are expensive are mainly full of kids from private schools and the elite students at them are also from public schools who are given full scholarships to raise the student academic profile.
Italy, Japan, South Korea and Turkey are examples.
> Italy, Japan, South Korea and Turkey are examples.
These are also countries with high prevalence of private, after school tutoring (i.e. so-called shadow education). For example, as of 2013 more than 40% of Italian secondary students attend private lessons. See Table 1 at
https://journals.sagepub.com/doi/full/10.1177/20965311198901...
And in Turkey this figure was nearly 60% in 2003, which may very well be even higher today (or at least # of hours is probably higher). See Figure 4.1 at
https://etd.ohiolink.edu/apexprod/rws_etd/send_file/send?acc...
Not going to bother posting figures for Japan and South Korea as I don't think anybody disputes the prevalence of tutoring in those or other East Asian countries.
If you want to prevent richer (either financially or in parental involvement) kids from getting ahead, you literally need to _suppress_ their ability to capitalize on those assets. Such as China is doing by regulating tutoring companies, or woke school boards are doing in various U.S. cities and universities. Of course, arguably you're merely hiding the underlying inequalities. But I guess this is just Goodhart's law in action, combined with pursuing the path of least resistance. That is, far easier to pursue a metric reflecting a consequence of economic inequality rather than metrics more directly measuring that inequality (e.g. Gini coefficient), let alone addressing the underlying causes themselves.
Japan’s public > private dynamic is further fueled by the fact that public high schools are entirely admissions test based, and students may choose which school they attend. It is also non-compulsory (although >90% of students attend high school).
You also only get 1 shot at applying to a public high school. So students tend to self-select (with some not-so-subtle nudging by their teachers) on the tier of public school they apply to based on various assessments of their own academic prowess.
This creates a competitive environment for the top tier public high schools, which feature heavily subsidized and reduced tuition. Yes. You must pay for high school in Japan. And competitive high-performing high schoolers tend to apply for and get into competitive, top tier universities such as Tokyo Universities.
Parents and high performing middle school children understand this acutely. So naturally, an entire industry of cram schools emerges to prepare these precocious 14 year olds to apply to the best high schools in their region so they can go to the nation’s best university so they can work at the nation’s best companies and live a life of middle class comfort.
On the other hand, private schools largely will overlook test scores in exchange for the more expensive tuition collected. As far as I can remember, you were also only allowed to apply to one private. Not to say privates are bad, or that kids don’t want to go to them. But the high performers definitely were going for the publics with the private school being their backup.
As a result, Japanese middle school students are some of the most stressed out teenagers I’ve ever seen.
The Paul Graham twitter thread:
1) At any given university, the kids from public schools are smarter than the kids from private schools. This is a necessary consequence of private schools improving kids' admissions prospects, and parents wouldn't send their kids to private schools if they didn't.
2) So if you show up at college and find that you're behind the kids who went to better high schools, don't be discouraged. Your position may be worse, but your velocity is higher. By the end of college you'll be ahead of them.
I think Paul's argument is misleading. I say that as someone who went to both in high school. There are public schools that put out the best of the best (i.e magnets and magnet-like schools) and there there are private schools that are just holding pens for the spawns of the wealthy and a few talented athletes. But one thing that many private schools provide that most public schools don't is the academic agency for students to pick from a wide range of courses that match their ability and a standardized curriculum that pairs well with those choice in terms of thoroughness and rigor.
Most public high schools don't have AP or IB courses and the ones that do tend to have them in limited supply in the hard sciences ( i.e. AP Physics C, AP Computer Science A, etc). Pursuing some sense of rigor involves taking courses that seem difficult rather than taking them in the hopes of forming a more cohesive understanding for one's future pursuits. Self-teaching obviously helps quite a bit, but that doesn't aid most students in achieving a firm foundation for say, undergraduate research in Physics.
Paul Graham's assertion that 'parents send their kids to private schools to improve university admission standards' does not guarantee that 'kids from public schools are smarter than kids from private schools.'
Some private schools have high admission standards, and these have better teachers and more motivated students than comparative public schools in the area. The existence of these counter-examples shows that Graham's reasoning doesn't hold. He also assumes that the public school students have studied less material, or easier material, than private school students, and are more motivated to learn. This isn't guaranteed; certain public school programs (like IB, mentioned by another commenter here) meet or exceed the curriculum standards by other private schools. In short, the wording of the Tweets in their current form demonstrate weak reasoning.
However, to take a more charitable interpretation, it's true that other private schools let anyone in who can pay the tuition, while also having less dedicated teachers and unmotivated students compared to public schools (especially public schools with high admission standards). Using more precise language, it can be altered to roughly: "motivated but less-privileged students can outcompete less motivated, more-privileged students in university."
This is a more interesting assertion, but it also depends on the amount of privilege (e.g. some privileged students get an apartment with near-zero commute to the university, lots of cash, the best tutors, etc.). Life isn't truly meritocratic, but you try to do the best you can with the cards you're dealt.
> _Paul Graham's assertion that 'parents send their kids to private schools to improve university admission standards' does not guarantee that 'kids from public schools are smarter than kids from private schools.'_
You are correct. However, it _does_ mathematically guarantee that 'kids from public schools are smarter than kids from private schools _at any given university_ (except, maybe, one of HYPMCS)'
In any competitive market, from the perspective of a market actor, independent selected-for variables will appear to be inversely correlated.
That's correct! Thanks for your comment; I wrote the original to practice logical thinking, and your feedback is helpful for learning.
So just survivorship bias for public schooled students? I feel like this only exacerbates how public schooling is failing
I guess I should be happy if this is true, considering my kids go to a public school, but there are just so many unstated assumptions to make the argument work that I could almost imagine it saying "...and for simplicity, we assume that students are spherical objects in a vacuum."
E.g., colleges know which public/private school each student is from. It's not a blind selection process.
I guess? I feel like this is one of those instances where it feels intuitively correct but probably isn't. Ignoring that, intelligence is only one dimension of success as well. I know plenty of wickedly smart people who flounder and do nothing because they'd rather stew in mediocrity than ever put themselves in a situation where they might fail. Seems like some unnecessary flavour text around what is essentially: "Do your thing and don't worry about everyone else."
>I feel like this is one of those instances where it feels intuitively correct but probably isn't.
I get this feeling reading a lot of 'insightful' Tweets, especially written by Paul Graham. His essays have been influential and useful, but his Tweets were disappointing (flawed arguments, emotional appeal) to the point where I unfollowed. It's just Twitter, but it does feel like a missed opportunity for a truly insightful feed.
Citation SORELY needed.
PG is saying it logically _has_ to be true as "a necessary consequence of private schools improving kids' admissions prospects", which subsequently has to be true because "parents wouldn't send their kids to private schools if they didn't [improve kids' admissions prospects]."
I personally think this is an interesting thought experiment—albeit also one with a lot of flaws. What if the kids from private school are smarter because the private schools have better instruction? Or what if only smarter kids are _accepted_ into private school, as private schools have a (frequently competitive) admissions process, whereas public schools have to take everyone.
Both of those could well be true. There’s just no data provided either way. Or any data about what cohort from either a public or private school actually goes to any particular university. It’s just annoying that the author comes out with a grand theory and provides nothing to back it up.
You don't need data to prove that the number of Nouveau apples someone purchases will go down if I, the sole farmer and distributor, raise the price. The application of an economic law to a new problem doesn't require the law to be re-proved.
The only evidence PG needs is that people continue to spend money on private schools in order to increase university admissions. That serves as evidence that private schooling increases acceptance rates in competitive admissions. It is a variable that schools select for.
When selections are made based on multiple criteria, and the market is mutually competitive, it follows that within each selection set (except sometimes those at the very, _very_ top) those criteria will be negatively correlated.
Because academic performance and private schooling are both criteria that universities use in selecting admitted students, those two traits _will_ be negatively correlated within each student body, i.e., "at any given university."
https://erikbern.com/2020/01/13/how-to-hire-smarter-than-the...
> The only evidence PG needs is that people continue to spend money on private schools in order to increase university admissions.
No, he also needs to prove that the people from those private schools aren't _in fact_ smarter than people from public schools who attend the same universities. Especially due to affirmative action, which has immense impacts on how people are selected to attend college.
We know people who send their kids to private schools tend to 1) have more money, 2) be more involved in their kids education, 3) more commonly have two-parent households. In short, they tend to be more successful and lots of that success can come from intelligence, however defined.
I am the first to poo-poo the academic establishment, but PG's reasoning is shoddy and I would be highly surprised if any data showed this to be true.
Please don't argue just to argue. PG is applying the very well known phenomenon of collider bias, a/k/a Berkson's Paradox. I have linked to a very thorough but approachable explanation of this phenomenon in a similar context (applicants for jobs rather than university).
This phenomenon is extremely well-established and PG's reasoning is flawless. His premise may or may not be accurate — perhaps private schooling doesn't make an applicant more attractive — but he offered his justification for that assumption.
If you have evidence to the contrary, please share it. If you can disprove collider bias, _please_ share it, and be sure to let HN know when you get your economics or biostatistics Ph.D. for that dissertation, because it would be pretty groundbreaking.
I graduated from a public high school and a public university, so I'm going to point out a seeming inconsistency with what you wrote and the web page that was linked to.
After checking out the Wikipedia page on Berkson's Paradox, I'm not as sure as I was about my original reasons for disagreeing with you and Graham, but I found a thread I want to pull on.
On the Wikipedia page, in discussing a hypothetical situation where the quality of burgers and fries may be positively correlated, but may also be observed to be negatively correlated due to the paradox, it says:
"...because they would likely not eat anywhere where both were bad, they fail to allow for the large number of restaurants in this category which would weaken or even flip the correlation"
I want to call your attention to the phrase "which would weaken or even flip the correlation". This appears to logically imply that the positive correlation _might not_ be eliminated or flipped, depending on specific circumstances.
Therefore your comments such as "[academic performance and private schooling] _will_ be negatively correlated within each student body" (emphasis mine), seem overstated or overconfident.
My gut reaction at first was simply that there is an implicit ceteris paribus that's taking too much load here; people aren't assigned to public and private schools at random, and I think that's essentially the same thing as abernard1 was expressing.
However, I can't say I completely intuitively understand the paradox, so apart from that, I question whether you are really stating the same principle as is generally accepted.
You wrote "you don't need data" and then asked for contrary evidence. If in fact, data is not required, then how can there be contrary evidence?
>> _because they would likely not eat anywhere where both were bad, they fail to allow for the large number of restaurants in this category which would weaken or even flip the correlation_
> _This appears to logically imply that the positive correlation might not be eliminated or flipped, depending on specific circumstances._
No. The negative correlation.
The text you quoted is explaining _why_ the negative correlation is observed, even though the underlying population data might show a weaker negative correlation or even a positive correlation.
Berkson's paradox is essentially a sampling error. Read it as "anecdotes have sampling bias when competitive market conditions force a trade off in qualities observed anecdotally." The only evidence necessary to show it applies is that there's an economic trade-off: Google hiring people based on experience and coding competitions, singles choosing dates based on personableness and attractiveness, schools admitting students based on educational pedigree and academic performance.
> _You wrote "you don't need data" and then asked for contrary evidence._
That isn't remotely what happened. I believe you may be conflating my discussion of apples with my discussion of schools. I encourage you to try to say only true things about others' positions, please.
>The text you quoted is explaining why the negative correlation is observed
Yes, and it is clearly describing the _possibility_ it shows up, which is inconsistent with it _always_ being the case. I'm trying to condense my point as much as I can.
>The only evidence necessary to show it applies is that there's an economic trade-off
This looks to me like essentially the same statement as I was referring to - "you don't need data". There is an economic tradeoff, and that's all we need to know.
I am _not_ telling you it's a fallacy to reject the possibility of contrary evidence. I'm saying that I doubt the principle is actually that powerful because of the qualified language I've mentioned already.
>That isn't remotely what happened
From my point of view, you just made the same claim again, so let's avoid the meta-argument and not start rehashing who said what. It only makes misunderstandings worse.
> _Yes, and it is clearly describing the possibility it shows up_
No. It is clearly describing the opposite: the possibility that the underlying situation shows an actually weaker, non-existent, or even inverse correlation.
The effect shows up when there's actually a negative correlation. It shows up when there's no correlation. It shows up when the actual correlation is a positive one.
So long as there's a trade-off of perceived positive qualities in a competitive market, it _always_ shows up.
>So long as there's a trade-off of perceived positive qualities in a competitive market, it always shows up.
>The effect shows up when there's actually a negative correlation. It shows up when there's no correlation. It shows up when the actual correlation is a positive one.
By "the effect" do you mean there is always an apparent negative correlation? Or are you including an effect that merely weakens a positive one?
Do you qualify your statements with the two events being independent?
> It’s just annoying that the author comes out with a grand theory and provides nothing to back it up.
Paul Graham has been doing this for 15 years.
I assume "smarter" is using the super crude definition "better grades"?
IMSA, a private boarding school in Illinois: "Two thirds of our students go on to earn a PhD. Our grads also created YouTube."
My school: "The guy who sang Chocolate Rain on YouTube went here!"
broke: This is an unsupported assertion.
woke: Why does PG want it to be true? What is he attempting to communicate and to whom?
Oddly enough this is the first time I agree with PG.
This is, I think, a misapplication of Berkson's paradox.
The assumption is that, all else equal, a university is less likely to accept a public school student.
But there's so many confounders here that Berkson's paradox is unlikely to follow trivially. This _might_ be true, but I expect it's more true for (and as a consequence of) parental income/legacyness than than preference for private schools.
In the USA a large percentage of private schools (possibly a slight majority) are religious and/or ideological in nature. Having experienced this first hand I would challenge everyone to question the assumption that private schools provide a better education. Some do, some don't. Sometimes the "product" being sold is purely satisfying a parental desire for control or safety (real or perceived).