College Admissions Statistics Class of 2020: Early Decision & Early Action Acceptance Rates

@spayurpets YTAR shows you the stark differences in colleges that you would usually consider to be in the same league. For example, everybody usually considers HYPMS very similar in terms of selectivity, but the YTAR shows that Stanford and Harvard are really in a very different league, with ratios almost 2 times that of Yale, Princeton and MIT and almost 5 times that of Cornell. That tells you how much more desirable these two institutions are in the eyes of the applicants compared to all other universities.

Now obviously you could look at the Number of applications, Admit rate and yield rate separately and probably guess that, but the YTAR brings it all together in a simple and expressive way.

@citivas has already addressed your other question. As an example, BYU has an incredibly high yield rate of over 75% but its YTAR is 1.59

I don’t think this is a measure of selectivity, exactly - really more of a measure of prestige. I think it becomes less useful at lower admit rate levels, though, because a small percentage change in the number of applicants will reduce the admit rate by a high (and increasing) percentage amount, causing the YATR to go parabolic.

For example, according to the numbers above, Stanford admitted 2,063 out of 43,977 applicants for a 4.7% admit rate and will have an 87.3% yield, assuming they’re looking to enroll a class of 1,800 (and leaving aside any waitlist effects). Therefore, its YATR is 18.6. If, however, Stanford’s number of applications rises next year by 2.3% (to 45,000) and they still accept the same number (2,063), the YATR rises to 19.1. If applications rise 4.6% (to 46,000), the YATR rises to 19.5.

Cornell, by contrast, had a similar number of applications to Stanford this year (44,966) but a much higher admit rate (14.0%). Its projected yield is 52.2% (assuming a desired class of 3,275), and its YATR is therefore 3.7. If its applications rose 4.5% (to 47,000), though, and acceptances and yield remained the same, the YATR would only rise to 3.9.

YTAR has the same algebraic inputs - Applications, Admits, Acceptances. YTAR = Apps x Acceptances / (Admits^2)

It may help differentiate the tail as the Admit rate approaches 0% and the Yield rate approaches 100% at the most selective / prestigious places. H and S at 18-20 is actually very impressive already. Assume 3,000,000 graduating students in the US all apply to a college with 2,000 places and, as it’s so desirable, it admits only 2,000 (assume 100% yield). Then, the YTAR is 1,500. This is a theoretical maximum, and I assume only US students apply. (Note that for bigger or smaller schools, the maximum could vary between 150 and 15000 for class sizes of 20,000 and 200 respectively). With only 30,000 applicants, the 1,500 YTAR comes down to 15. With 60,000 applicants, this would be YTAR=30. In practical terms, it will be hard to see >30 unless the whole world starts to apply here.

But if students from China and other developing countries keep coming, we could see H and S - or others - getting 180,000 applications one day for the same 1,800 class size, for a YTAR of 100. Again, assuming a 100% yield.

There is also a question of social waste. We are seeing this already at YTAR levels we observe today. Do we want students, college counselors, and admissions offices to be spending hours on what will end up being unsuccessful matches? For the benefit of making the class and the fit just that little bit better? (No reason to doubt that many qualified applicants are rejected at HYPSM etc.) The social price to pay on 60,000 applicants going for 2,000 places (from the YTAR=30 example) would be 58,000 unsuccessful applications multiplied by X man-hours on each, multiplied by the number of colleges witnessing this phenomenon. Plus much emotional damage.

@VeryLuckyParent interesting approach. We also saw in earlier discussions (#356-#366) that the raw Admit / Yield rate already hides a lot of the true selectivity and desirability, for which we need to look carefully at ED vs EA, how much of the class is admitted in ED, etc. Some of the Ivies at RD nowadays get a 50% Yield or less, and losing out to other Ivies or MIT, Stanford, Chicago, Duke, etc.

Adding CMC to these tables:

Ranked Admissions Data + Yield
Class of 2020: Top 20 Liberal Arts Colleges
*

Total Applications:

1.Colby College 9,822
2.Middlebury College 8,820
3.Amherst College 8,397
4.Pomona College 8,104
5.Swarthmore College 7,717
6.Grinnell College 7,368
7.Vassar College 7,306
8.Williams College 6,982
9.Bowdoin College 6,788
10.Carleton College 6,500
11.Claremont McKenna College 6,342
11.Hamilton College 5,434
12.Wellesley College 4,888
13.Harvey Mudd College 4,180

RD Acceptance Rate:

1.Pomona College RD ~566 out of 7190 (~7.9%)
2.Claremont McKenna College 594 out of 6342 (9.4%)
3.Harvey Mudd College RD 421 out of 3716 (11.3%)
4.Bowdoin College RD 687 out of 5918 (11.6%)
5.Amherst College RD 969 out of 7943 (12.2%)
6.Swarthmore College ED/RD 963 out of 7,717 (12.5%)
7.Middlebury College RD 1042 out of 7866 (14.2%)
8.Williams College RD 960 out of 6397 (15.0%)
9.Colby College ED/RD ~1720 out of 9822 (17.5%)
10.Grinnell College ED/RD ~1326 out of 7368 (~18%)
11.Carleton College ED1/ED2/RD ~1430 out of ~6500 (~22%)
12.Hamilton College ED/RD 1317 out of 5434 (24.2%)
13.Vassar College ED1/ED2/RD 1943 out of 7306 (26.6%)
14.Wellesley College ED/RD ~1368 out of 4888 (~28%)

Early Acceptance Rate:

1.Harvey Mudd College ED1/ED2 ~77 out of 464 (16.6%)
2.Pomona College ED1/ED2 ~177 out of 914 (19.4%)
3.Bowdoin College ED1/ED2 284 out of 870 (32.6%)
4.Amherst College ED 180 out of 454 (39.6%)
5.Middlebury College ED1/ED2 398 out of 954 (41.7%)
6.Williams College ED 246 out of 585 (42.1%)
7.Davidson College ED 207 out of 458 (45.2%)

Total Acceptance Rate:

1.Pomona College ED1/ED2/RD 743 out of 8104 (9.2%)
2.Claremont McKenna College ED1/ED2/RD 594 out of 6342 (9.4%)
3.Harvey Mudd College ED1/ED2/RD 498 out of 4180 (11.9%)
4.Swarthmore College ED/RD 963 out of 7717 (12.5%)
5.Amherst College ED/RD 1149 out of 8397 (13.7%)
6.Bowdoin College ED1/ED2/RD 971 out of 6788 (14.3%)
7.Middlebury College ED1/ED2/RD 1440 out of 8820 (16.3%)
8.Williams College ED/RD 1206 out of 6982 (17.3%)
9.Colby College ED/RD 1720 out of 9822 (17.5%)
10.Grinnell College ED/RD 1326 out of 7368 (18%)
11.Carleton College ED1/ED2/RD 1430 out of 6500 (22%)
12.Hamilton College ED/RD 1317 out of 5434 (24.2%)
13.Vassar College ED1/ED2/RD 1943 out of 7306 (26.6%)
14.Wellesley College ED/RD 1368 out of 4888 (28%)

Yield:

1.Claremont McKenna College 58.1% (345)
2.Pomona College 55.2% (410)
3.Bowdoin College 51.5% (500)
4.Middlebury College 47.6% (685)
5.Williams College 45.6% (550)
6.Swarthmore College 43.6% (420)
7.Wellesley College 43.5% (595)
8.Amherst College 41.1% (472)
9.Harvey Mudd College 40.2% (200)
10.Carleton College 37.3% (495)
11.Hamilton College 36.4% (480)
12.Vassar College 34.4% (668)
13.Grinnell College 33.2% (440)
14.Colby College 28.2% (485)

*Excluding, for insufficient data: Colgate, Davidson, Haverford, Smith, US Naval Academy, Washington & Lee, Wesleyan.

@DeepBlue86 Exactly. YTAR shows you how hard it is going to be for even a school like Cornell to catch-up to Stanford any time soon in terms of prestige and selectivity. Universities that are playing this game are looking at some very sad odds, just like applicants are. Personally I also think YTAR could be a leading predictor of which schools are more likely to increase undergraduate enrollment in the near future, because schools with high YTAR relative to their peers are in a better position to increase incoming class size without sacrificing student selectivity and prestige.

@veryLuckyParent I’m sorry but that’s a load of hooey. As @bronze2 rightly pointed out, you’re basically taking the square of the same bunch of algebraic inputs, and the product is meaningless.

@VeryLuckyParent YTAR = (#Enrolled)x(#Applied) / (#Accepted)^2 Expressed this way, does it make any sense what’s in the numerator or why you would square the #Accepted in the denominator? Garbage in garbage out.

@spayurpets regarding your question on
quotex(#Applied) / (#Accepted)^2

[/quote]

That is the same as Yield rate/Admit rate with those two terms expanded. I will let you work out the details by expanding the formula for Yield rate and admit rate.

As I have said before, YTAR is not a new metric, it is a summary metric that pulls together a bunch of metrics in an expressive and interesting way in my opinion. I did not invent this metric! It is a standard metric used in many admission discussions to compare colleges.

I understand that, I was just noting that when you unpack it, you realize how none of the inputs make any sense. What’s happening is that you are blowing up the #accepted in the denominator by squaring it, so you exaggerate the gaps between the schools. It’s not telling you anything about the relative position of each school.

Well, @spayurpets and @VeryLuckyParent, I wouldn’t go so far as to say it’s “hooey” or “garbage”, but I think YTAR has some issues, for the following reasons:

First, I’m not sure what YTAR explains, or how one should interpret it. Stanford’s YTAR is 18.6. Harvard’s is 15.8. Yale’s is 11.0. Cornell’s is 3.7. Does this mean that Stanford is roughly 18% more prestigious (whatever that means) than Harvard, or 1.8x as prestigious as Yale, or more than 5x as prestigious as Cornell?

Second (as previously noted), at the high end, small movements in total applications lead to exaggerated movements in YTAR. Stanford and Cornell get a very similar number of applications (44 to 45 thousand), but an incremental 2,000 applications (less than 5%) increases Stanford’s YTAR by roughly 5x the amount it increases Cornell’s. If both Stanford and Cornell increase their total applications by similar amounts and percentages in a given year, can we really say that Stanford became significantly more prestigious relative to Cornell that year?

Third, as has also been noted, the numerator and denominator are both “game-able” in ways that can have complex effects. Ginning up applications from unqualified students will decrease the admit rate without a direct effect on yield, of course. I think the yield numbers are particularly suspect, though, because (i) they are undoubtedly pumped up at schools with ED programs (particularly if the school fills an outsize proportion of the class early); (ii) they rely on assumptions about targeted class size which aren’t necessarily disclosed by the schools, may change year over year, and therefore are often no more than estimates; and (iii) they don’t take account of how actively/aggressively the schools manage their waitlists, or if they have Z-lists.

Fourth, certain schools have “market niches” that further reduce comparability if one is trying to measure relative prestige. Harvard is the biggest worldwide brand name in education; its total applications and yield are undoubtedly inflated as a result. I think most people would agree that Stanford is the best university west of the Mississippi and has less regional competition than the Ivies, it’s also identified closely with Silicon Valley and engineering, and all this is no doubt pumping up its numbers.

On the other hand, Princeton and Yale, which are of similar size to Harvard and Stanford at the undergraduate level and (according to US News, at least) are of at least equal quality to them, don’t have these differentiators. Accordingly, they get 29 and 31 thousand apps instead of 39 or 44 thousand like Harvard and Stanford, and their yields are 69 percent instead of 82 to 87 percent - even though all four schools have admit rates between 4.7 and 6.5 percent (which makes them fundamentally indistinguishable in terms of difficulty of entry).

This discussion prompted me to graph a scatterplot of admit rate (on the vertical axis) vs. yield at the 10 schools for which @spayurpets provided numbers (the Ivies, Stanford and MIT) and derive a best-fit line (for the nerds among you, y=0.4034e^-2.48x, R-squared = 0.7217). I think this provides some interesting results: (i) Cornell, Penn and MIT seem to have high admit rates relative to their yields; (ii) Brown, Columbia, Princeton and Yale seem to have low admit rates relative to their yields (although the differences from the best-fit line are much less than in (i)); and (iii) Dartmouth, Harvard and Stanford are right where the best-fit line suggests they should be. I’d invite others to chime in if they have thoughts on this.

On a lighter note … how much is climate and location a factor in yields and admit rate, i.e. number of apps? Stanford and Pomona/Claremont McKenna are top in their respective categories (by a healthy margin in the latter case). There is less choice west of the Mississippi to be sure, but it is a huge plus to be able to enjoy the outdoors year round. It’s not all brand and silicon valley entrepreneurship - climate is rarely mentioned! If these schools were in Alaska or North Dakota, the brand might not carry nearly as well.

@spayurpets @VeryLuckyParent Is there some way to consider what percent of the class is admitted ED in comparing yield to admit ratios? It seems like yield on your RD pool is more impressive than yield on the overall class. Especially for the LAC’s which have ED and not EA.

@Corinthian It is only possible for the ED schools that break out ED, and it’s a lot of math. We can work on it.

The problem with yield I am discovering is that the class size is easily manipulable. I think several schools make up an unreasonably high “target” class size that overstates yield. They probably figure that come fall everyone will forget that they overestimated the enrollment by 50-75 students.

Ranked Yield for ED Schools, Total and RD (Estimated Class Size):

1.University of Pennsylvania 66.8% (2445) (RD 1110=47.7%)
2.Brown University 56.9% (1660) (RD 991=44.0%)
3.Pomona College 55.2% (410) (RD 237=41.1%)
4.Dartmouth College 54.0%(1175) (RD 681=40.5%)
5.Cornell University 52.2% (3275)(RD 1937=39.2%)
6.Bowdoin College 51.5% (500) (RD 216=31.4%)
7.Duke University 51.4% (1705) (RD 892=35.1%)
8.Northwestern University 51.3% (1925)(RD 864=32.1%)
9.Vanderbilt University 48.1% (1600)(RD 800=31.7%)
10.Middlebury College 47.6% (685)(RD 287=27.5%)
11.Tufts University 46.8% (1325)(RD 662=30.5%)
12.Williams College 45.6% (550)(RD 304=31.7%)
13.Johns Hopkins University 41.6% (1300)(RD 716=28.2%)
14.Amherst College 41.1% (472)(RD 292=30.1%)
15.Harvey Mudd College 40.2% (200) (RD 123=29.2%)
16.Scripps College 32.8% (245)(RD 132=20.9%)
17.George Washington University 22.6% (2474)(RD 1633=16.1%)

@bronze2, I’m sure you’re right. If you want to attend a tippy-top university located where it hasn’t snowed in at least 40 years, you’ve got exactly one choice (or arguably two, if you fit the profile for Caltech and consider it a tippy-top). That fact undoubtedly enhances Stanford’s numbers.

I have been wondering why it might be the case that (based on my chart of admit rates and yields at the Ivies, Stanford and MIT), Cornell, Penn and MIT might have high admit rates relative to what their yields suggest. If this is in fact true, I have some possible explanations:

  1. Cornell and Penn have ED programs, which enhances their yield relative to EA schools, all else being equal (and particularly if they weight the class toward students they accept ED).
  2. Cornell and Penn lose admits to HYPS, LACs and state schools, so have to have a relatively higher admit rate for the RD round than HYPS (whose relatively fewer losses of admits are, I would guess, mostly to each other).
  3. Unlike HYPS, where there’s one way in, there are various specialized undergraduate schools at Cornell and Penn. It could be that some of them have high admit rates while those they admit tend to enroll in greater numbers. Maybe that’s the case at the Cornell School of Hotel Administration, or the College of Agriculture and Life Sciences, for example. I wonder how the admit rate and yield at Wharton compare to those at Penn’s College of Arts & Sciences.
  4. Similarly, I’m going to guess that because MIT is itself relatively specialized, the applicant pool is reduced and the school probably has to admit a greater percentage of applicants, but because it’s arguably at the top of its category, it has a high yield.

From the RD yield chart, it is easy to see that the Ivies still get a higher student preference to their peer institutions, out of proportion to their rankings; i.e., Brown, Cornell, Dartmouth and Penn have significantly higher RD yields than Duke, Northwestern, Johns Hopkins and Vanderbilt. The Ivy League is still a powerful brand.

I would also draw the conclusion that Bigger is better. Universities have significantly better yields than LACs.

@DeepBlue86 @bronze2 I don’t think it’s so much the weather as distance that explains the Pomona preference. Many students don’t want to have to travel more than two hours to get home, and that’s going to limit how many Californians and other West-of-the-Mississippians are going to go to Williams and Amherst. In this regard, Pomona has no LAC competitors, other than the other Claremont colleges and possibly Colorado College and Occidental.

Hoping that @nostalgicwisdom checks in on this thread and may have some insight on the Pomona applicant pool. We know the geographic make up of the admitted students but not the applicant pool. https://www.pomona.edu/sites/default/files/pomona-college-admissions-2015-16-profile.pdf

@spayurpets #398 Stanford Daily reported class size of 1730 following interview with dean of admissions soon after RD. I think somewhere else, a class size of 1800 may have been mentioned and used. With quite a few looking to expand class size in next few years, movements can change yield by a few %age points

“Z lists” can also make a difference. Harvard acknowledges Z-listing 30-50 students a year. If that group has a yield of close to 100%, which presumably it does if its members are agreeing to take a year off in order to go to Harvard, it could noticeably increase yield on a projected class of 1,675.