Question about "yield protection"

But aren’t REA kids signaling that Harvard is a first choice by virtue of using that restriction there and not somewhat else? I would expect this group would have a very high yield at any school that offered it. Sure, a kid might also apply to Stanford or Princeton in the RD round, but presumably they chose Harvard for the restricted round because it was their first choice.

It’d be interesting to see how Princeton’s decision not to participate in the SCEA/REA round in 2020 affects the SCEA/REA yield that year of the other schools, particularly Harvard and Yale (Stanford may be less affected).

1 Like

UCs certainly look at yield for the purpose of yield prediction, even though they do not practice yield protection. However, they have made high profile mistakes in yield prediction in the past (either over or under enrolling a campus in a given year).

It would not be unreasonable to assume that every individual applicant who is admitted or at the margin of admission is assigned a predicted yield value based on all of the applicant’s characteristics that have been found that correlate to yield at that college. Such characteristics include (but may not be limited to) academic stats, admissions reading scores, intended major, race/ethnicity, financial situation, level of interest indicators, and probably other things. Besides offering admission, the college can also influence an applicant’s predicted yield in some ways – the most obvious is offering scholarships, honors or other perks, or preferentially packaged financial aid, but some other types of sales and marketing methods can also be used (e.g. how admitted student visits are handled).

Here’s another way to protect yield - if they perceive you lack interest, they yank you out of the calculation. Hadn’t thought of that one!!!

From @DooniesG

CWRU withdraws application for unsubscribing from their emails - College Admissions - College Confidential Forums

3 Likes

I also saw that and was surprised by it. If we believe they are keen to lower their acceptance rate, wouldn’t it be a better move to not say anything and just reject the applicant?

4 Likes

Crazy. A Lehigh AO said something a bit similar: they track how often applicants access their portal…and for the ones that never did/never set it up, let’s just say they didn’t receive good news on decision date.

2 Likes

I see Case’s response as a bit different than Lehigh’s. Case bombarded us–kid and me–with emails. There is a very long time between when an application is sent and a decision is released, and it is stressful for some kids to get so many reminders during this waiting period.

4 Likes

This (checking whether applicant used the portal) was mentioned in an article back in 2013 about Lehigh admissions:

2 Likes

I believe applications that are withdrawn are still included in the denominator for the calculation of acceptance rate.

I think that both Case Western’s and Lehigh’s practices noted here are ridiculous. There are many individuals who think it’s actually healthy to step away when awaiting a decision when there is nothing you can do but wait. So Case Western wants to bombard people with emails which can definitely start to feel like spam, even if it’s your top choice (keeping a clean inbox is a preference of many). And Lehigh wants people to be obsessively checking their portal to see if there’s any news. Both institutions just went down in my esteem this evening.

3 Likes
However, the lawsuit also found that REA applicants tend to get a significant boost in chance of admission.

If an applicant is applying to Harvard REA because they expect a “boost” compared to applying RD, that applicant is going to be disappointed.

While it is nominally true that in a typical year, the Harvard REA admission rate is about twice the overall admission rate, REA is also the stage of the admissions process where Harvard athletes are recruited for Ivy League sports. Recruited athletes should always be considered a separate case from all other categories, including others of special institutional interest, because it is the only category where Harvard actively seeks out (aka, recruits) individuals in a formal process where the applicants are vetted for suitability, desire to attend Harvard, and likelihood of acceptance of an admission offer even before an application is submitted. All Ivy League schools follow a generally agreed upon process for recruiting athletes, a process that is also governed by NCAA D1 rules.

A more useful comparison would be one between Harvard REA applicants with recruited athletes removed and the overall applicant population. Even then, the applicant profiles differ (that is, the REA pool is usually stronger across all evaluation categories than the general pool.)

Harvard states, as do many other similarly selective schools with Early Action programs, that for any given applicant, applying REA confers no admissions advantage compared to that same applicant applying RD.

Early Decisions, however, are a different case and many schools that offer ED are explicit or nearly explicit that admissions preference is given to those who apply by those deadlines. EDs are structured to both improve prediction values and to improve yield.

2 Likes

During the lawsuit period, the admit rates for non-ALDC applicants (excludes both recruited athletes and legacies) were as follows.

Non-ALDC REA Applicants – 14.5% Admit Rate
Non-ALDC RD Applicants – 3.2% Admit Rate

After controlling for the ratings of the applicant, URM status, and dozens of other factors; the Plantiff’s model found an average of a ~4x (odds ratio) increased chance of admission associated with applying REA instead of RD for non-ALDC hooked applicants (excludes athletes and legacies). I believe REA admissions advantage is statistically significant on the 99.99999999999999999999999999999999999999% level, for the baseline sample with ALDC hooks removed – not likely to just be a coincidence.

In the most recent prior to COVID, the admit rates for the full class were as follows. Unfortunately only numbers for the full class are available. If you exclude athletes and other ALDC hooks, the admit rate for both REA and RD would no doubt be significantly lower than the overall totals.

Overall REA Admit Rate – 13.44% (56% of class filled REA)
Overall RD Admit Rate – 2.79% (44% of class filled RD)

I’m fascinated that you have computed that given a pool of REA Harvard applicants, taking out recruited athletes, Harvard lineage candidates, those on the Dean’s Interest List, and children of faculty and staff (ALDC in Arcidiacono parlance) yields a subgroup with a 14.5% Harvard admissions rate after factoring out “dozens” of other factors like the applicants ratings (!).

This is very good news for unhooked applicants to Harvard.

The admit rates are listed in Table A.4 from the Arcidiacono analysis. This table separates admit rates for “regular admits” and “special circumstances” = ALDC, showing both REA and RD. I took the average of the 2016-19 years for which the RD/REA admission model was in place. In more recent years admit rates have been lower, so I expect the 14.5% admit rate for unhooked applying REA has decreased, as has the 3.2% admit rate for unhooked applying RD.

The admit rates above are the actual admit rates. There is no controls for applicant ratings or other factors. The regression analysis with controls for reader ratings of applicants, hooks, and various other factors is separate and in a different table towards the end of the document.

Harvard is far from the only school that gives an admission advantage for applying via a restrictive early policy that is associated with a high yield, which includes ED. It just happens to be the one for which specific numbers are available, due to the lawsuit. At other colleges, I expect there is a good degree of variation in degree of preference for students applying via a restrictive early policy over students applying via regular decisions, ranging from near 0 benefit to much larger than the numbers suggested in the Harvard lawsuit.

Ah…that makes sense, since that table has no control for applicant ratings, which lie at the crux of the early stages of the Harvard admissions process aside from recruited athletes. So, these data can be true and it can also be true that for any given applicant who is not in a category of interest, applying REA confers no significant marginal advantage. That’s what Harvard, MIT, et al say, and I think they tell the truth. Why would they misrepresent this when other institutions are open about ED being a marginal advantage in admissions?

Harvard experimented with eliminating REA, and decided it needed to return to re-capture desirable candidates for competitive reasons. So, in a general sense, this was about yield. Having returned to REA, however, I don’t think Harvard’s fear of positive yields is enough to confer any meaningful advantage to candidates applying early vs regular. If an applicant won’t be admitted RD, they probably won’t be admitted REA.

The admit rates in table A.4 show the actual admit rates for the full class. The admit rate table is not a regression analysis with controls. Separately there is a regression analysis with controls in table B.7.1R that does control for applicant ratings, hooks, stats, race, and dozens of other criteria… enough criteria to explain the majority of variance in admission decisions. This regression analysis found the average of 4x admission boost (odds ratio) for applying REA among non-ALDC applicants, with full controls That is if you compare applicants the same hook status, same reader ratings, same …, you’d expect the one who applies REA to have a 4x greater chance of admission than the one who applies RD on average. The sample size was many thousands of applicants, so the standard error was low enough to be for this result to reach 99.99999999999999999999999999999999999999% statistical significance. For statistical purposes, near certain that applying REA offered a significant average advantage to non-ALDC (not athlete, legacy, …) during the lawsuit period.

2 Likes

So it’s a win-win for them (lower admit rate, middle finger to the applicant).

If it’s a win-win for one side, it must be a lose-lose for the other.

1 Like

Congratulations to your son for his admission to ND. If you applied early to ND, I assume that was his first choice? If so, why spend anytime worrying about a deferral to Case?

1 Like

My belief is no decision, no # in the calc. So it won’t lower their acceptance rate but would support the yield.

I don’t know what happens but that’s my guess because they withdrew, they weren’t rejected.