This came up in the decision topic and I disagreed with the advice of senior members. I lay out a toy problem here to illustrate what a school like Penn State (large flagship state school that is frequently used as a safety school for top-admits) stats may look like.
Please consider this dataset as an example of why you would use enrolled over admitted data. The numbers are fake, but I think it illustrates the effect.
Let’s consider this purely fictional set of numbers to illustrate the difference between enrolled and admitted profiles given that Penn State’s profiles for each are measurably different.
For ease of numbers we will consider three sets of students:
- Group 1 members have 4.0 unweighted GPAs
- Group 2 members have 3.5 unweighted GPAs
- Group 3 members have 3.0 unweighted GPAs
Assumptions:
- Penn State yields 10% of 4.0 GPA admitted students (meaning 1 out of 10 enroll historically)
- Penn State yields 50% of 3.5 GPA admitted students
- Penn State yields 100% of 3.0 GPA admitted students
- Penn State wants an enrolled class of 100
The fake 2023 applied class looks like this:
Group 1: 100 applicants
Group 2: 100 applicants
Group 3: 100 applicants
- So, Penn State auto-admits all of Group 1, knowing it will get 10 students enrolled (10% yield). They now need 90 more to fill a class of 100.
- That means Penn State admits wants to also admit 100% of Group 2, knowing they will enroll 50 students of the 100 admitted (50% yield). They now need 40 more to fill a class of 100.
- That leaves the 3rd group. Penn State needs another 40 students, so they admit 40 out of the 100 in Group 3 because that group has a 100% yield.
This results in 240 admitted students and 100 enrolled.
Here are the numbers broken down for the Class of 2023:
Admitted GPA (240 students):
Admit rate: 80% (240/300)
Admit 75%: 4.0 (60th person is in Group 1)
Admit 25%: 3.5 (180th person is in Group 2)
Enrolled GPA (100 students):
Enroll 75%: 3.5 (25th person is in Group 2)
Enroll 25%: 3.0 (75th person is in Group 3)
This is how the numbers between the admitted and the enrolled differ. See the GPA drop between the two? This type of drop can be seen in Penn State’s reported numbers. So, if you are an applicant and have a 3.0 - what should you look at for guidance? Admitted stats or enrolled stats?
Clearly, it is enrolled stats. 40% of those kids in my example got accepted in that group, not much less than 25% as suggested by the admitted stats. Why? Because those top-admits don’t yield to Penn State as they do to Harvard and Stanford.
Penn State uses the admit stats as a marketing tool because it looks better. That’s why the CDS uses enrolled stats. That paints the picture of who enrolls at a school. Schools know their yield rates in each tier of students. Throwing a pile of low yielding top-admits into the mix doesn’t significantly affect the bubble tier’s ACTUAL chances. But the reported numbers can make it look much worse than it is.
This example is extreme, but it shows how the numbers work in this scenario or low yielding top admits.
I’m open to any counter-arguments.