Data10, I know you interview for Stanford whose legacy policy includes all schools in the university. If that’s the case most of university legacy applicants would not have parents connected with regional interview committees who may consist mainly of college alums. The selection bias I was referring to may not be there then. I would be interested to see legacy boost at Stanford.
The Harvard lawsuit data analysis focused largely on this one page data summary sheet—those 4 numerical assignments. But from what I saw, the admission file could be 3-5 pages long, and deliberation contained far more things than those four numbers in the table. Though the 4 numbers are highly correlated with the admission outcome they may not be the deciding factor in many cases.
I expect most legacy applicants at other colleges are also not connected with regional interview committees. Interviewing is voluntary, and most alumni do not volunteer. I also don’t think Stanford is the only selective college where alumni parents may strongly encourage their not well qualified kids to apply, perhaps due to not realizing that admissions has become far more selective since when they attended. While alumni on this forum tend to be quite knowledgeable about college admission, there are also alumni who don’t spend much time reviewing how college admission has changed.
In any case, I don’t doubt that alumni applicants tend to be stronger than non-alumni on average. Using some specific numbers from the Harvard lawsuit and assuming LDC are representative of L (the vast majority of LDCs are L), then some example numbers are below.
White ORM Applicants (the vast majority of legacies are White)
Good (1-2) Academic Rating – 54% of LDC applicants, 45% of non-ALDC applicants
Average (3) Academic Rating – 40% of LDC applicants, 45% of non-ALDC applicants
Near Auto Reject (4) Academic Rating – 5% of LDC applicants, 10% of non-ALDC applicants
Asian ORM Applicants
Good (1-2) Academic Rating – 63% of LDC applicants, 60% of non-ALDC applicants
Average (3) Academic Rating – 32% of LDC applicants, 32% of non-ALDC applicants
Near Auto Reject (4) Academic Rating – 5% of LDC applicants, 8% of non-ALDC applicants
I listed 4 numbers as an example, rather than exhaustive list of all controls. There are more like 10+ numerical ratings (GC ratings, LOR ratings, interview ratings, …) that were included in the analyses, Stats including GPA and multiple scores were included, as were hooks, ECs, and dozens of other controls. It’s not the complete file, but it’s enough to explain the majority of variance in admission decisions. The Plantiff’s r (correlation) for model 6 was 81% (first analysis that includes athletes). Harvard’s expert achieved a similar r. While not perfect, I think a model that has an 81% correlation with admission decisions is strong enough to draw some meaningful conclusions. Apparently both sides of the lawsuit did as well, as the model was a key tool used to make their case in trying to win the lawsuit.
H’s class is 14% legacy (230 kids), defined as parents or grandparents who attended the college. If legacy acceptance rate is 20% (vs 4% for non-legacy) with a 95% yield, then each year the number of legacy applicants is 1216 (230/0.95/0.2=1216). With 1650 grads each year and each alum having two kids and four grandkids, you would have about 9900 legacies on average each year. Out of 9900 legacies only 1216 apply, a mere 12.3%; that is not a picture of many encouraging parents and grandparents. I suspect there is some self selection going on. In fact, if 60% legacies apply they would have ended up having the same acceptance rate as the non-legacies.
Harvard defines legacy as parent attended Harvard, not grandparent. In the older lawsuit sample 13.9% of admitted students were parent legacies. Given the higher yield for legacies, it would probably be 14-15% of matriculating students. In the most recent pre-COVID freshman survey (class of 2023), 16.8% of the class said they were parent legacies. The median of 5 pre-most recent pre-COVID classes (2019 to 2023) was also 16.8% parent legacies.
In the lawsuit sample, the acceptance rate was 33.6% for legacy vs 5.9% for non-legacy. As discussed in the previously linked “divergent path of legacy and athlete” study, the admit rate for legacies tends to remain similar over time since there is a limited supply of legacies and portion of class that is legacies remains similar over time. So I think a better estimate for recent admit rate might be ~30% admit rate for legacies compared to ~4% for non-legacies. These percentages are also more in line with the Princeton numbers above.
If you eliminate grandkids (grandparents not counted as legacies for admission purposes), then you need to cut your total estimated supply of legacies by 1/3, which would increase the legacy application rate to 37%. However, you also need to decrease further below 37% for a variety of reasons including the higher legacy admit rate listed above. For simplicity, I’ll use a 30% application rate. A comparison between this application rate for legacies and the application rate for non-legacies is below:
Legacies – Hypothetically 30% Application Rate (see above)
Non-Legacies – ~30k Domestic Harvard Non-Legacy Applicants prior to COVID / ~4 million US HS graduates = <1% Application Rate
Based on the application rates above, which group appears to have more self-selection – legacies or non-legacies? Also note that there are numerous reasons why a legacy kid might not apply to Harvard besides just their parents convincing them they have little chance of being accepted. Many kids use their ED/REA/SCEA single application on a different first choice college and get accepted, many kids want to stay in a different region of country (or world), many kids prioritize something in colleges that Harvard does poorly, many kids want to attend the same nearby college as their friends/boyfriend/girlfriend/community members, many kids intentionally want to pursue a different path than parents, some families prefer lower cost options (after scholarships) or cannot afford Harvard due to unique financial circumstances, etc. For example, when I was a kid, I was a prospective EE major. At the time Harvard didn’t offer much of an EE program, so it was off the list. It wouldn’t have mattered if my parents were legacies, there would be no point to applying to a college that offered little in the field that I wanted to study.
That said, for Harvard we don’t need to make crude estimates and make rough guess at numbers. The lawsuit sample tells us how many legacies applied; how the applying legacies admission ratings, stats, hooks, and other factors compare to non-legacies; and tells us how strong the remaining legacy boost is after controlling for the above. And all sides of the lawsuit + the Harvard Office of Internal Research all agree that the remaining legacy boost is quite strong.
The legacy boost may be stronger at some colleges than at others. None of those colleges that offer legacy prefererences is transparent about its practice. But how strong (or weak) that boost is also seems to depend on what the applicants intend to study in college. A STEM applicant to a top private is likely to get much smaller boost because it’s easier to distinguish stronger STEM students from less-strong students, and the farther out to the right side of the distribution (which many applicants to the top privates tend to be in) the greater that distinction. On the other hand, similarly strong non-STEM applicants tend to be less distinguishable from each other and their legacy statuses may play a greater role.
As an example, Stanford, with its larger portion of so-called “techies”, is likely to offer smaller legacy boost, on average, than Harvard, with its larger portion of “fuzzies” (again borrowing a term from Stanford). In the same vein, the “techiest” colleges (Caltech or MIT) offer no legacy boost (by their own policies). Although HMC states in its CDS that it considers legacy status, an applicant’s legacy status is likely to play only a very minor role, if any. And I’m sure there’re counterexamples or exceptions because of the very nature of holistic college admissions.
With well advertised legacy preference that rate seems low. I wonder how many of those 30% happen to be active alums who are donors and interviewers. I imagine there is a lot of overlap.
And of course, if donating time and money translating into a big boost then this would be something not captured by the data analysis.
My experience is not with H. And perhaps the boost there is strong.
I do know that the vast number of ordinary legacies at my alma mater are rejected (and ended up at other colleges with sub 10% acceptance rates). These are very strong students with stellar academic records.
Every candidate whose parents donated significantly or volunteered a tremendous amount was accepted. They weren’t terrible, but they were generally as a group were not as strong as the group above.
Future candidates should come to their own conclusions.
Around here, the two very public legacy/athlete admits to Stanford were both double legacies, both parents athletes, both students athletes too. Jessica Elway (father John a football player, mother Janet a swimmer) and Christian McCaffrey, (father a football player, mother a volleyball player). Are they counted as legacy admits? Twice?
Interesting theory, but do you have any Stanford data backing up the supposition that legacy provides less of a boost at Stanford. Or, if it does, this is because it is more “techie” and less “fuzzy?”
I’m not sure that using CalTech supports your theory, given that CalTech is now test blind and thus may be relying more on “fuzzy” factors.
As for HMC, there has been some mixed messaging, and it is still not clear to me whether or not HMC gives legacies preferential treatment, given that HMC was required by California law to report if they “provide any manner of preferential treatment in admission to applicants on the basis of their relationships to donors or alumni” and HMC reported it did not. Also, as @Data10 pointed out in a previous thread, the President of the HMC indicated hat HMC doesn’t really use legacy admissions. Here is the strange quote:
So I don’t think we know whether HMC provides a legacy boos or not, or in what circumstances it does (if it does), or if that has anything to do with it being a “techie” school.
Harvey Mudd claimed that they do give advantage to legacies, so they did not submit any specific numbers to CA government. Stanford said that they admitted 302 students who met the legacy preference criteria in the bill (276/302 enrolled), and 34 non-legacy admits who met the philanthropy preference criteria in the bill (28/34 enrolled).
What we do know is that a much larger portion (about 2x?) of a typical Harvard class consist of legacies than that of a typical Stanford class, consistently, even though Stanford uses a much broader definition for its legacy admissions. Do we know how legacy statuses factor into each of these two colleges admission decisions? No, we don’t.
I wouldn’t associate “techie-ness” with standardized tests. For Caltech, those scores weren’t that meaningful anyway, even before the announcement about their new test blind policy.
Yes. However, given HMC core requirements, I don’t expect legacy preference (or even donor preference, which HMC supposedly practices as well) played a significant role in its admissions.
My post above mentions 276/1698 matriculating students in Stanford’s class of 2023 (pre-COVID) meet the AB 697 definition of legacy advantage, which is 16.3% of class. This is similar to the percentages reported in Harvard freshman survey, and slightly above Harvard lawsuit sample. However, as you note, Stanford uses a broader definition of legacy than does Harvard.
I think we know a lot more about legacy admissions at Harvard than nearly any other college due to the lawsuit. However, non-Harvard colleges rarely volunteer information about degree of legacy preference and have motivation to downplay legacy preferences in public statements, so there is very limited information. Some people believe that legacy generally provides a big boost at certain specific non-Harvard colleges; some believe hat legacy provides a big boost only in a specific, limited circumstances; and some believe that legacy generally doesn’t have much influence. At all legacy preference colleges besides Harvard, there isn’t much we can do to prove which of these 3 options is most accurate.
Based on differences in admit rate, I think it is likely that legacies provide a lower average boost at Stanford than Harvard and Princeton. However, the difference in admit rate could also more relate to Stanford’s broader definition of legacy… maybe the additional categories average less weight. It’s largely a matter of speculation.
I see @Data10 already covered this, but here is the backup regarding Harvard (14.6% legacies in Class of 2023) and Stanford (16.2% legacies in Class of 2023)
The same California report indicated that they don’t give donor preference, so more ambiguity.
Regardless, I don’t understand why HMC’s core requirements would lead you to believe that legacy preferences couldn’t play a role in admissions. Is it your hypothesis that with “legacy” admits would not be qualified to meet the demands of the core curriculum?
My recollection of 2x rate at Harvard (compared to Stanford) should be about their respective legacy admit rates (not class proportions). My mistake.
This article above revealed some of the admission practices at Stanford. With respect to legacy preferences, it mentioned that applications from legacy applicants were read by two separate readers (unlike those from non-legacy applicants), and that the “percentage of alumni children admitted to Stanford is roughly three times the overall percentage of acceptance”. At Harvard, that ratio is about 6, twice as high.
Do you really believe that a significant percentage of HMC alums don’t think their own kids could hack the core requirements at HMC? Even if true, some alums would very likely think their kids could hack it, right? So I don’t understand how this supports your hypothesis. Or, after seeing the Stanford numbers, have you rejected your hypothesis?
@ucbalumnus, I think that’s posted above, but thanks for reposting.
From what I’ve observed, MOST alums of MOST colleges have a reasonable understanding of what would work for their kid or not.
You graduated from a tech university and your kid wants to major in graphic design and fine arts. You get it. You graduated from a music conservatory and although your kid is “very musical” and in fact, may be a fine musician, you know that it is or isn’t the right environment for your kids talents and aspirations.
I don’t think this is about who can or cannot “hack” the requirements- it’s a lot more nuanced than that. The “Hail Mary Pass” applications colleges get from legacies are when the parents are either clueless about what their kid wants (it happens) or have not kept up with what it takes to get admitted these days (it happens).