Hacker News new | past | comments | ask | show | jobs | submit login
Legacy and Athlete Preferences at Harvard [pdf] (duke.edu)
77 points by ibobev 23 days ago | hide | past | web | favorite | 50 comments

Table 5 and 6.

If you're a random (non-ALDC) African American in the 30-40% (4th decile) range for academics you have the same chance of getting in as an Asian American in the top 10%. They each have about a 12% chance.

> Our model of admissions shows that roughly three quarters of white ALDC admits would have been rejected if they had been treated as white non-ALDCs.

If we're looking at how group membership affects acceptance rates, why not ask how many white ALDC admits would get in if treated as African American non-ALDCs? Why not ask how many Asians would get in if treated like Hispanics? Why not ask how many African Americans would get in if treated like Asians? The whole paper looks like a reflection of the cultural value that it's a good thing to go after rich white people, a bad thing to go after anybody else.

The interesting comparison is in Table 6 Panel B: White LDC vs African American non-LDC. For the top 60% academically just being black carries about as much advantage as being in the elite 4% (roughly) of white people who have some form of deep connections to the school. If you're in the bottom 40% you're better off being white with connections.

Socially we are conditioned to not question legacy admissions, while being very critical of affirmative action. Your comment is an example of this.

Consider: Table 6, which you cited, shows that legacy dunces (3rd decile) have a better chance of getting into Harvard than academic toppers (top decile) who aren't legacies. Yet you chose to focus on the much smaller gap between admit rates of African Americans and Asian Americans.

Table 6.

The numbers you pointed out:

White LDC 3rd Decile: 16.67% White non-ALDC 10th Decile: 15.27%

The "much smaller gap" that I chose to focus on:

African American non-ALDC 4th Decile: 12.76% Asian American non-ALDC 10th Decile: 12.69%

3rd vs. 10th decile in one case is worth an academic paper with 3 authors collaborating from 3 different universities. Plus their associations with NBER and IZA.

4th vs 10th decile in the other case is something I was socially conditioned to see and is not worth focusing on at all.

I think that if the study's authors had completely messed up a spreadsheet and replaced the numbers for the White LDC/White non-ALDC columns with the numbers from the African American non-ALDC/Asian American non-ALDC columns, you would suddenly find those numbers to be troubling.

Doesn't legacy admissions have a racial impact since the racial mix of students has been changing so rapidly? That is, white applicants benefit the most simply because the % of white Harvard alums (relative to all alums) is greater than the % of white applicants (relative to all applicants)...but in a steady state system where race was the same over a couple generations, that impact would cease. I'm not a fan of legacy admissions in general but if you wanted to keep it in place, the fairest way might be to silo its impact within each racial group, so that a white legacy has an advantage over non-legacy whites but no advantage over non-legacy applicants of other races.

I guess I don't agree with your idea that this is a reflection of the idea that it's a good thing to go after rich white people. The problem with analyzing college admissions to this point is that there's no one else to go after. By percentage, who else can you rag on? There's just no other demographic that's been going to college as long. In 2016 at least 57% of college students described themselves as white, which has fallen since the 1976 level of 84% [1]. And college has always been for the well-off, not the poor.

If you want to you could figure out how many white ALDC admits would get in if treated as African American non-ALDCs; I'd encourage you to then normalize across high school quality and expense, too. And looking at the lot of Asian-Americans in US college admissions is certainly important right now. For a quick peek, just look at Caltech's admits, since they have historically ignored legacy and race unlike everyone else.

But again, who else are you going to rag on if you're going to investigate the effect of legacy admissions and prof's children admissions? Due to the historic legal structure of the US it's just going to be really white. Fortunately for us ladies, men have daughters at a relatively high rate and we're allowed to go to college now so we can now benefit from unfair advantages too. (Again, it's instructive to look at Caltech, which did not admit women until the early 1970s and famously did not hire Emmy Noether because she was female. The percentage of women students has risen steadily as access to math and science education has become more available to the girls of the college-going classes of the US.)

[1] https://nces.ed.gov/fastfacts/display.asp?id=98

> look at Caltech...[who] famously did not hire Emmy Noether because she was female

What? I'm having trouble finding a reference. I know the story that she faced discrimination and Hilbert vouched for her, but that was in Germany. I didn't know there was any from Caltech.

Why is it assumed that we have to rag on people? If we are ragging on people, what standards are we using? Obviously it's still ok to get artificial help to the top if you're in some groups, but not in others. That means it's ok to be discriminated against if you're in some groups, too.

Speaking of Caltech, female applicants have 3x the chance of getting into Caltech as male applicants[0]. It's almost certain that qualified males aren't getting in because they're males. That was the problem with Emmy Noether: she was qualified but couldn't get in because she was female. If the discrimination right now is against the males, why are we talking about Emmy Noether being forced to audit classes in 1903 on the other side of the world?

If we assume "equality" in the groups of male and female students applying to Caltech for 2016-2017, meaning they both should have the same acceptance rate, then 134 seats were taken from the guys because they're guys and given to the girls because they're girls. Maybe you can argue the girls only apply if they're really serious, but 3x?

[0] - https://www.parchment.com/c/college/college-178-Caltech.html...

edit: here's a better data source: https://nces.ed.gov/collegenavigator/?q=caltech&s=CA&l=93&id...

It shows 3x admissions advantage for females with over 2x enrollment advantage. Everybody's saying they don't do affirmative action but it's hard to believe at least for gender. But the racial demographics are extremely different from Harvard's so maybe the female applicant pool really is 2x as likely to be qualified. I need more data.

I was an undergraduate at Caltech and helped with some admissions work. There is no bias towards admitting females. The student body is extremely small (200-250/year) and the student pool extremely self-selective. The 4 year graduation rate is troublingly low enough (people will transfer or drop out) that admissions will not admit people if they can’t meet the academic bar (and the curriculum is structured so that the first 1.5 years regardless of major require the same classes). They do however emphasize top-of-funnel outreach for women to try to get more to apply.

> There is no bias towards admitting females./ They do however emphasize top-of-funnel outreach for women to try to get more to apply.

The 266/267 ratio of acceptance letters is probably too good to be true, it's hard to believe they could so perfectly calibrate this outreach program.

If the school is spending money recruiting females and not males it demonstrates they're more valued even before starting the application process. That's a bias, and I know at my school that bias which started before the application process extended all the way through the 4 year experience and even into the hiring process where companies have internal numbers to hit.

I have a lot of respect for Caltech and they deserve credit for not taking part in the widespread anti-Asian discrimination.

There’s too much focus on credentialism. Personally, I think it’s fine to have mediocre rich people at Harvard.

If anything, it’s better for society to get smart people in other colleges. Too much brainpower is wasted on the investment banking and consulting mills that the Ivy League is sending people into these days.

Top universities have top funding and research that if it meets top talent can be extremely useful for the world.

Usually the figures about admissions are undergrad focused. Are Ivy League schools doing legacy admissions for graduate research?

The critical factor when considering these numbers is schools chose metrics that fit their desires not the other way around.

The direct weights given to each category is only part of the story, it’s also where the breaks exist within each category. Allow more people to hit rank 1 in a category directly reduces the impact of that category. Basically, does an academic 5 start at 27 and below ACT or 25 etc etc.

THIS. schools have mandates on revenue, for T1 research place like Harvard 1/3 is from tuition and 1/3 from alum donation, so that’s wealthy kids, and incentive for their parents to donate. Overtime schools realize there are polite metric to measure for wealth and likelihood of parents donating, and use these metrics to hit their revenue mandate. Also recruiters come to Ivies looking for cultural fit, so you have to serve up a certain kind of product (people) to appease your recruiters

I don’t know about the whole University but only ~25% of the operating budget of the Harvard school that includes Harvard College comes from students (tuition, room, & board) and that has been on a decline as more and more of admitted students receive financial aid.

Donations used directly for operations is way, way less than 1/3rd. A little over half the operating budget comes from interest on the endowment, I don’t know how to figure how much of that can be attributed to donations rather than investments, patents, etc.

Ah my mistake, I meant to say proceeds on endowment. The development office is always hounding people for more gifts though

Dumb question- why is 1/3 from donation when the endowment is so massive?

I did my undergrad at a school with very relaxed admissions requirements and never got into any schools that use “holistic admission” like Harvard (or for that matter Duke, where I got rejected) does.

It’s clear to me that any attempt to make these admissions criteria is doomed to fail because *there is no good definition of “qualified”, and there can’t be. I’d wager if you dropped any 75th percentile student from NC State at MIT they’d probably perform well enough to graduate. Not 5.0, but with enough grit they’d do it.

I think more people should accept that all selective admissions is at some level social engineering. Making it more “equal” is a noble goal - but it’s still social engineering. These efforts would be better spent improving institutions that lead to actual class mobility rather than arguing whether 20% or 30% of Asian applicants to Harvard should get in for nebulous reasons.

I've worked with both MIT and NC State grads and in all 3 cases MIT was far ahead. Now whether it was the CS program that was better or the raw smarts of the grads or both, I don't know.

I think that's missing the point because the question is whether the NC student would do well at MIT, not whether the result of the program is better.

Can you expound on this? In what way were they "far ahead"? Under which circumstances?

I went to NC State, for the record. Am I just "far behind"?

Sure, MIT grads knew many general CS and systems concepts more in depth than their NCSU counterparts. They would also finish their artifacts way faster than the NCSU grads. This is just anecdata and may or may not reflect the average grad quality from both these schools.

Well, thanks for your candor in looking down on 99.7% of Americans. Usually people don’t say the quiet part loud.

You asked my experience and I told you with a disclaimer. You clearly aren't interested in it, rather than some platitudes like "everyone is the same". Again, I have no personal grudge against you or NCSU or other state schools. For that matter NCSU has a good systems program from what I hear.

Your original reply was not really a material response to my point about social engineering at all - you just said folks at MIT are superior in a direct comparison. It's hard not to take away from that a direct attack on average folks like me, but I will say this - I'm not at all surprised.

I qualified my response saying it's anecdata. Your responses clearly try to undermine the top engineering institutions, I provided a counter.

Given a choice would you not have picked MIT over NCSU ? The CS program there is quite amazing as well as the opportunity to meet other smart people and minor in a plethora of courses.

I don’t think you understand as someone that isn’t a member of the cognitive elite I would have never been given the choice.

The question prefaces by giving you that choice.

Getting into MIT is hard; graduating is not particularly difficult.

It’s like this at most elite institutions.

NC State has a graduation rate of 75%. Duke (where I was rejected for being an inferior candidate) has one that’s 95%.

The percentage means nothing. It could be that some NC State's candidates may fail at both NCSU and Duke. And most Duke will pass at both universities.

There was no reason to single out whiteness among ALDCs (athletes, legacies, Dean's interest list, and children of staff). This seems like it's trying to lie with statistics.

They're overwhelmingly white, and the lie is pretending otherwise.

Yea, but the point is that its not necessarily causal. It would be like looking at people who died of cancer and determining weight-loss as one the best correlated culprits to ID it. It would probably catch a ton of people who were really sick, but it seems to ignore some of the deeper casualties.

The white thing certainly is problematic culturally, but it's not REALLY the problem its more of a symptom of previous power choosing future power.

I agree with you, overall. So why dance around it? In this country (the US), white people by law got to win more often than others. In India and China and other places other rules were in place. Every place has its rules, and in most, as you put it, previous power chooses future power. With our weird "one drop" laws in the US affecting everything from voting to banking access to property rights to mortgage rates to schooling, race and ancestry are inextricably tied to that in the US. Ruby Bridges only had her 65th birthday this Sept 8. She's younger than my parents.

I think it is utterly fascinating to watch the world change and watch morals and mores change, but still see power reconfiguring, like mercury falling into a groove, to maintain power and bend law to its service! And the naive/idealistic among us say oh, we'd never do things like our ignorant and unenlightened ancestors. We don't see race. We don't see color. We're a pure meritocracy, with advancement ruled only by your access to the best schools ;)

Except in the US, in spite of the historically large majority white demographic makeup of the country, we have done more to be welcoming to other groups in recent history than virtually any other country on Earth. Look at a chart listing salary in America by country of origin. That chart tells the story of a country that gives people from all over the world opportunity to succeed.

> I think it is utterly fascinating to watch the world change and watch morals and mores change, but still see power reconfiguring, like mercury falling into a groove, to maintain power and bend law to its service!

Absolutely beautifully put.

So ... what do you want? To re-initialize Harvard with a 100% randomization of admissions for 10-15 years?

Would help if the abstract didn’t uses phrasing like “benefits whites.”

It doesn’t benefit whites, it benefits kids with rich parents. Similarly, Affirmative Action benefits kids with lesser academics, but explicitly based on their race.

Perhaps if they want to draw apples-to-oranges comparisons they would do well to point this out.

> Would help if the abstract didn’t uses phrasing like “benefits whites.”

You have made two factual misrepresentations about the content of the paper in this thread so far. That doesn't seem like discussing in good faith.

For reference, the abstract of the paper is:

> The lawsuit Students For Fair Admissions v. Harvard University provided an un- precedented look at how an elite school makes admissions decisions. Using publicly released reports, we examine the preferences Harvard gives for recruited athletes, legacies, those on the dean’s interest list, and children of faculty and staff (ALDCs). Among white admits, over 43% are ALDC. Among admits who are African American, Asian American, and Hispanic, the share is less than 16% each. Our model of admissions shows that roughly three quarters of white ALDC admits would have been rejected if they had been treated as white non-ALDCs. Removing preferences for athletes and legacies would significantly alter the racial distribution of admitted students, with the share of white admits falling and all other groups rising or remaining unchanged.

The paper does not use the phrase “benefits whites.” Interpreting the paper as such is a racist interpretation. You would do well to read the paper.

pg 2: "legacy preferences will tend to benefit white applicants relative to other racial groups.

pg 34: "Each of the ALDC preferences primarily benefit white students."

I'm sort of troubled by the fast & loose nature of mieseratte's statements; in such a simple matter one might as well be accurate about the location of the statements referenced, and WillPostForFood is correct that the statements that contain "benefit white" students/applicants are not in the abstract but later. I also think that there's a bit of subtle misrepresentation when this discussion is being framed as "benefiting whites". There are many white people in the US who do not have access to college, do not have access to good public education, do not have the financial means to become legacy students, etc. In the sentences from the article, there's some set theory going on: among admitted students, a larger fraction of white students were benefited by ALDC preferences. Among applicants, a larger fraction of white students got the benefit of these admissions. This is a race thing, sure, among people already applying for college -- which is already a substantial filter with its own complexity.

Many colleges care about legacy admissions because they cement a family (and their donations and alumni networks and job help and etc) to a college more securely than anything else. It is also true that this perpetuates the patterns of the past. One of those patterns in the US is that white people got to go to college and others did not. White people should not get too sensitive about this. It's just true.

It's refreshing to see a statistical analysis that is free from the whole "affirmation action" rhetoric that's posted in articles about this case. Though I always had a general sense that ALDC's were admitted at a higher rate, I never knew that it was this high.

Did you not read the paper? They make direct comparison to AA in the first paragraph.

No it does not. I just reread the opening paragraph three times and there is no comparison being made to AA. Did you not read the paper?

From the second paragraph of the introduction:

"The ongoing debate about the use of affirmative action in college admissions has also drawn attention to the impact that other admissions preferences have on the racial composition of college students"

Observing that lots of people are concerned about affirmative action is not the same as advocacy for affirmative action. Observing and even criticizing discrimination in favor of white applicants is also not the same as advocacy for affirmative action.

This is a great example of beating numbers to fit the author's narrative. Take this statement:

> The advantages for athletes are especially large, with an average admit rate for recruited athletes of 86%

Athletes at Harvard are recruited by coaches who look for students who are good athletes and also will be admitted. They know what the admissions committee looks for and won't recruit any student who doesn't have a strong chance of getting in.

I'm sure recruited athletes have an advantage, but saying it's "especially large" because a pool of applicants hand picked by coaches have a high admit rate doesn't make any sense.

For the record, I believe colleges like Harvard should stop giving a boost to legacies and athletes. But we need to a have fact-based discussion about it. Papers like this pretending to be research that are really op eds, don't help anything.

I'm sure recruited athletes have an advantage, but saying it's "especially large" because a pool of applicants hand picked by coaches have a high admit rate doesn't make any sense.

They are doing more analysis than you suggest. They are looking at Athlete admits, and comparing their scores in other categories like academics, and noting they are much lower than other admits. I.e., the paper is not just looking at admit rate.

Recruited athlete admits are universally weaker than non-ALDC admits on these ratings. This is not surprising, given that we know athletes are stronger on the athletic rating. But for some race and rating combinations, the differences are striking. At most, 28% of white athlete admits receive a 2 or higher on the academic rating. In contrast, 89% of white non-ALDC admits receive a 2 or higher on the academic rating. 78% of Asian American non-ALDC admits receive a 2 or higher on the extracurricular rating, while at most 12% of admitted Asian American athletes receive a similarly high extracurricular rating.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact