They do not necessarily do it all the same way. For example, if USNWR#19 UCLA does holistic review in a way similar to the way UCB does it as described in the Hout report, the procedure is (in greatly simplified form) as follows:
- Each application is read and scored by two readers. If the scores are too different a third senior reader reads the application and rescores it.
- Within major/division, applications are ranked by the reading score averages, and a cutoff for admission is determined to produce the desired number of admits (presumably based on yield estimates). Obviously, those above the cutoff are admitted, while those below are rejected (or waitlisted).
- Tie-breaking procedures are used when a group of same reading score average applications straddles the cutoff.
Note that this procedure differs from the assumed procedure that everyone is writing about here in that:
A. All applications go past step 1 to step 2, rather than some being rejected at step 1.
B. Step 2 does not involve a central admissions committee rereading each application. Considering the number of applications that UCLA gets every year, and the number of applicants admitted from them, having a central admissions committee reread each application, even after aggressive winnowing, would not scale.
Of course, when leaving the top 20 of USNWR, many colleges will get significant differentiation among applicants by academic stats alone, so there is less need for them to use holistic reading to differentiate between applicants crammed up to the academic stat ceilings. So they may use point systems, and/or have automatic admission for high stat applicants, etc…