###
Introducing Kai!

Your College Confidential guide bot.

Kai can provide tips and support as you research and apply to colleges, and explore majors and careers.

Stay on top of the information you need to navigate the admissions process amid the COVID-19 pandemic. We've got articles, videos and forum discussions that provide answers to all of your test prep, admissions and college search questions.
Visit our COVID-19 resource page.

Your College Confidential guide bot.

Kai can provide tips and support as you research and apply to colleges, and explore majors and careers.

SatchelSF
1372 replies13 threads Senior Member

Just a little tidbit to emerge from the Harvard Asian discrimination documents that runs counter to the seemingly prevailing wisdom on CC that tippy-top elite schools could "fill their classes" with perfect stats kids: for the class of 2019, which is the only data I have been able to find so far, in the entire admit pool, there were fewer than 1000 applicants **at most** who could have presented even the first component of "perfect stats":

SAT 1600 - 361 applicants (~1.0% of applicants)

ACT 36C - 625 applicants (~1.7% of applicants)

I assume that some portion of these are overlaps (maybe 10%, maybe a little more?), but even assuming each score in uniquely assigned to an applicant, Harvard couldn't even fill half of its admits with perfect scoring kids.

Adding constraints of perfect GPA will whittle that number down further, although perfect scaled GPAs are much more common (by a factor of at least 8x), and as high school grading is so inflated on average it can be assumed that a sizable percentage of perfect scorers also present perfect GPA (maybe 50-60%, maybe more?).

There's no need to get to perfect SAT subject test scores or APs, it's pretty clear that if we add those conditions there are not enough applicants in the entire pool to even fill a small size lecture class (that's just math), even if Harvard took every one of them.

No need to go into the whole "it's holistic" shtick/spiel - I get it,**believe me**, and probably a lot better than most people who tout it - I'm just offering this score tidbit for discussion.

As noted above, I assume the scores are single sitting. But Harvard says it only codes the "superscore." If these 986 scores in fact represent superscores, the actual number of single sitting perfect kids will be lower.

If these are (as I believe) single sitting scores,**why is Harvard tracking these**? Hmmm. I wonder if in its elusively mysterious criteria for Academic 1s on their effectively 1-4 rating scale (of which there are only about 100 applicants a year iirc), perfect single sitting scores are a component?

A lot more goes into admissions decisions than scores ("it's holistic" - again, I get it). But it's worth noting that admit rates for perfect score kids (SAT or ACT, not even including GPA) appear to hover somewhere above 30%, more than**twice** the admit rate of kids with scores even 50 SAT points lower (or equivalent ACT points - so much for the "threshold" idea?), and of course very far above average admit rates. No doubt perfect GPA+ACT/SAT rates are substantially higher than 30% - probably at least 40-50% I'd guess.

(Scores in themselves of course are not that important; the high admit rates for perfect scorers reflect the group's much higher average intelligence, which will be positively correlated with all the other factors that are important to schools; so, it's not that the scores get the kids in, it's more that the scores evidence their underlying excellence on many other important dimensions as well, again on average for the group.)

Data on number of perfect GPA, SAT and ACT applicants on p.12 here: https://projects.iq.harvard.edu/files/diverse-education/files/expert_report_rebuttal_as_filed_d._mass._14-cv-14176_dckt_000419_037_filed_2018-06-15.pdf

Data on admit rates by score on p.7 here: http://samv91khoyt2i553a2t1s05i-wpengine.netdna-ssl.com/wp-content/uploads/2018/06/Doc-421-145-Admissions-Part-II-Report.pdf

Note that these numbers come**directly from Harvard sources**, so they are as accurate as the competence of the adcoms will allow.

115 replies SAT 1600 - 361 applicants (~1.0% of applicants)

ACT 36C - 625 applicants (~1.7% of applicants)

I assume that some portion of these are overlaps (maybe 10%, maybe a little more?), but even assuming each score in uniquely assigned to an applicant, Harvard couldn't even fill half of its admits with perfect scoring kids.

Adding constraints of perfect GPA will whittle that number down further, although perfect scaled GPAs are much more common (by a factor of at least 8x), and as high school grading is so inflated on average it can be assumed that a sizable percentage of perfect scorers also present perfect GPA (maybe 50-60%, maybe more?).

There's no need to get to perfect SAT subject test scores or APs, it's pretty clear that if we add those conditions there are not enough applicants in the entire pool to even fill a small size lecture class (that's just math), even if Harvard took every one of them.

No need to go into the whole "it's holistic" shtick/spiel - I get it,

As noted above, I assume the scores are single sitting. But Harvard says it only codes the "superscore." If these 986 scores in fact represent superscores, the actual number of single sitting perfect kids will be lower.

If these are (as I believe) single sitting scores,

A lot more goes into admissions decisions than scores ("it's holistic" - again, I get it). But it's worth noting that admit rates for perfect score kids (SAT or ACT, not even including GPA) appear to hover somewhere above 30%, more than

(Scores in themselves of course are not that important; the high admit rates for perfect scorers reflect the group's much higher average intelligence, which will be positively correlated with all the other factors that are important to schools; so, it's not that the scores get the kids in, it's more that the scores evidence their underlying excellence on many other important dimensions as well, again on average for the group.)

Data on number of perfect GPA, SAT and ACT applicants on p.12 here: https://projects.iq.harvard.edu/files/diverse-education/files/expert_report_rebuttal_as_filed_d._mass._14-cv-14176_dckt_000419_037_filed_2018-06-15.pdf

Data on admit rates by score on p.7 here: http://samv91khoyt2i553a2t1s05i-wpengine.netdna-ssl.com/wp-content/uploads/2018/06/Doc-421-145-Admissions-Part-II-Report.pdf

Note that these numbers come

This discussion has been closed.

## Replies to: Perfect score + GPA applicants are actually pretty rare even at Harvard

Yes, it was out of 2400, but Harvard internally uses an average scale up to 80. I was just converting to put it into current terms, and yes iirc there were typically about 400-600 2400 scores per year in the data I have seen.

I have always suspected that schools are not being wholly truthful about superscoring, and I think this graph supports the idea that at least for some scores they do record the single sitting score. It would be difficult to manually look through all the data if there weren't a field somewhere that recorded the scores presented. Alternatively, of course, these perfects could be the result of superscoring, which would suggest that Harvard attracts a slightly lower percentage of perfect scoring applicants than the numbers indicate.

It does seem that Harvard is not quite as popular among perfect ACT scorers (only 625 out of about 2000 true 36 scorers) as among perfect SAT scorers (361 out of, say, 550).

Analogy: which is harder, earning A grades in a high school using ABCDF grades, or earning 100 grades in a high school using 0-100 numerical grades?

I’m going to guess that the ~100 academic “1”s include some perfect scorers, but that a lot of them are published authors/researchers or winners of national academic competitions who have less than a perfect score in the section of the SAT/ACT that doesn’t include their sphere of excellence.

This also accords with logic. For some people, getting a 1500 after a few summers of prep and a few actual test attempts is a challenge. On the other hand, a few kids can score 1500+ at 12 years old without even getting out of second gear. From which group would you expect to find a higher rate of Academic 1s when application time rolls around? Which group will on average have the time to get to, say, grad level math courses, or publishing research, or a novel while still in high school? Remember, within the group of 1600 scorers are a few kids who could score 1800+ if the test were made harder, although there are none of those in the 1550 scorer group by definition. Perhaps this phenomenon accounts for the apparent doubling of admit rates between SAT 1550 to 1600, for unhooked kids. That's quite a jump for 50 points, no? If these are one dimensional grade grubbers, it's surprising that Harvard's system doubles their share, no?

If you go through the academic stuff in the reports, it's pretty clear imo that the Harvard system is actually fairly simple. There is a mechanical SAT/ACT+GPA calculation that outputs an academic index, distributed uniformly across ten deciles. Despite the absence of fine granular data, it also seems pretty clear that the Academic Rating - which is really only available from 1 through 4 - is highly correlated with that index. There is some small adjustment for rigor I'd imagine for a few schools (consider that at a place like Davidson Academy, the bottom 10% will probably be as intelligent as the average top 10% at just about any other school), and these adjustments will certainly make a difference in say, pushing a kid from a 3+ to a 2-, or a 2 to a 2+, but in general given the size of the pool and the limited discretion that junior adcoms must have, it's probably pretty mechanical.

Academic 2s are basically 50% of the applicant pool (60% of Asians, 46% of whites basically gets you there), and that's where all the action is for unhooked kids. The Academic 3s and 4s are basically "no hopers" imo, without special hooks.

Take a look at p.15 in the doc linked below and particularly at the notes on academic decile-Academic Rating correlation (and instructive infographic). Given the constraints (no + or - data on the individual ratings, and as noted by me above the possibility that, say, 3+ => 2- or vice versa cannot be captured in the correlation), a 0.70 R^2 (implying ~0.84 correlation) is actually extremely high. There may be less to Harvard's system than meets the eye here. Except for the extraordinary candidates - and I agree with @collegedad13 here regarding faculty involvement because I know it to be true - most of the rest are likely simply mechanically assigned (with slight adjustments for some sending schools and reflecting some reader idiosyncratic factors including error). At least that's how it looks to me.

http://samv91khoyt2i553a2t1s05i-wpengine.netdna-ssl.com/wp-content/uploads/2018/06/Doc-421-145-Admissions-Part-II-Report.pdf

Not the strongest evidence, I admit. It may well be that the 361 perfect SAT scores are actually superscores, so the "true" single sitting number is somewhat lower, as I noted above.

I think that Harvard will not superscore the ACT across different sittings, but will consider the highest single sitting, so the ACT 36s represent single sittings.

The basic point remains that perfect stats kids are going to be incredibly rare, which of course is common sense. More interesting is trying to infer where the cutoffs for the top academic decile really fall. I was assuming around >1580 SAT/36 ACT + >3.9 GPA would be the approximate cutoff, but just eyeing the numbers, I think that is too high. Perhaps 1550/35 + 3.8 GPA gets you into the top 10% of the applicant pool, certainly into the top 20%. I'll try to look at that a little because it is an interesting question for potential applicants and imo more valuable than the "it's holistic/everyone has great stats" advice that's always offered on here.

We all know that Harvard's raw admit rate is very low (~5.0%), but most of the unhooked applicant pool obviously consists of "no hopers" based on academics alone, so the denominator needs to be adjusted by applicants who are trying to think about chances.

You can think about it in terms of combinatorics.

For the SAT, there is only

1 combothat nets a single sitting 1600 or 2400:800 + 800 (+ 800 for 2400 scale tests)

For the ACT, there are

15 combosfor a true composite 36 if my math is right:4 sections 36 -- 1 way

3 sections 36, 1 35 -- 4C1 = 4 ways

3 sections 36, 1 34 -- 4C1 = 4 ways

2 sections 36, 2 35 -- 4C2 = 6 ways

Maybe one of the compsci guys

girls- I've learned my lesson, lol - could confirm, it's been a long time since discrete math for me....Drew Gilpin Faust said valedictorians. I thought Fitzsimmons referred to "qualified" applicants.

@skieurope ?

Easier for a test taker who is top-end at both tests to score the maximum score, due to the coarser scale.

But, away from the top end of the scale, some test takers find the ACT easier, while others find the SAT easier.

Prevailing wisdom on the forums can get inaccurate, like the game of telephone.

A quote form the Harvard Statement of Material Facts is below. It states that 2+ academic rating includes some applicants with "perfect" grades and testing. And it implies being one of the rare few that get a 1 academic rating (fewer than 0.5% of applicants in lawsuit sample) involves criteria beyond just having perfect stats.

"An applicant receiving a “2+” academic rating is typically an applicant with perfect, or near-perfect, grades and testing, but no evidence of substantial scholarship or academic creativity....

In many circumstances, an applicant receiving a “1” academic rating has submitted academic work of some kind that is reviewed by a faculty member. "

I agree that the higher admit rate primarily relates to scores being correlated with other parts of the application rather than score itself. However, I doubt that 2400 scores typically have "much higher average intelligence" than the numerous other Harvard applicants who score a few points shy of 2400. Instead I'd expect the bulk of the perfect scores are persons who took the test several times, until they managed a 800 on each section at least once, which superscored to 2400 Superscoring would explain why the number of 2400's in Harvard's applicant pool is so large in relation to the total number scoring 2400 as listed by CollegeBoard.

The Arcidiacono regression coefficients suggest the primary driver in the academic component of admission is academic rating, rather than test scores. After controlling for academic rating, the additional contribution of AI was quite small. The small component that does remain may largely relate to +/- academic ratings distinction, which were not controlled for.

Of course academic stats are an important component of academic rating. The percentages in different academic deciles who received a 2 (or better) academic rating is below, as listed by Arcidiacono. There is clearly a correlation, but it's also clearly not a mechanical function of an AI threshold.

Top Decile -- 98%

2nd Decile -- 94%

3rd Decile -- 84%

4th Decile -- 70%

5th Decile -- 51%

6th Decile -- 26%

7th Decile -- 8%

8th Decile -- 1%

Academic rating explained 9% of variance in admissions decisions (Card model), and academic index explained ~half of variance in academic rating. Harvard OIR's model was able to explain a larger portion of variance in academic rating through grades and scores than Arcidiacono was able to in his full model that including several functions of academic index, gender, race, concentration, hook status and many other factors. Harvard's OIR model's regression separated grades and scores, which allows a greater relative weight onconverted GPA (or whatever GPA is listed in file) and lesser weight on scores; while Arcidiacono's model forced the AI score heavy weighting (2/3 of AI is scores), which may suggest Harvard is internally places a greater emphasis on grades.