Simulated Applicant Pool

In order to show you the view form the admissions office, I undertook an incredibly dull task -- I turned every applicant on a particular grid to an index number.  

The particular grid I chose was UCLA's, in the 2000 Official Guide.  I chose UCLA because there are no "minority" admits based on race to cloud the issue, and because the acceptance rate acts like that of a private school.  Normally, state schools have a higher yield at the top of the applicant pool, because of low tuitions and regional preferences.  Those trends don't apply to UCLA; Berkeley is a "better" state school that will attract the top end of UCLA's acceptances, and UCLA gets a lot of east coast applicants who will pay a high tuition wherever they go (since there are very few low-cost state law schools on the east coast).

Once I chose my school, I began number-crunching, as follows:

Each box on the applicant pool grid represents about 7 index points.  E.g., assuming a 10 to 1 index ratio, the box "160-164/3.5-3.74" covers index numbers from 195 (160 + 35) to 201 (164 + 37).  

I took this box on the UCLA grid and distributed the applicants over the 7 index points, assuming more people would have lower index numbers at this point in the range.  For the 336 applicants in the "160- 164 /3.5 -3.74" box I assigned them thus:

Then I distributed the 87 who were accepted, assuming a lower number as index number decreased, thus:

Finally, I asked myself, "How many of those accepted will attend UCLA?"  Given that UCLA accepted 840 people to fill 275 seats, we have to assume a lot of people who declined.  Who are they?  

• People at the top who got into a better school.  (In fact, I had to cut a lot more at the top than I had anticipated in order to get the yield down to 275).
• People in the middle who got into a lower schools with big scholarships;
• and people at the bottom who decided not to travel so far from home.  

Since UCLA's median index is about 199, I assumed that at index #'s of 200 and higher, folks are going elsewhere.  As the index number decreases, the yield increases; for applicants with an index # of 197, UCLA may be the best school they got into, increasing their likelihood of attending. So I assigned yield rates as follows:

 So combining those three steps, we get the following chart:

All of that work represents a single box of the 108 on UCLA's grid!  Now repeat 108 times, roughly 700 index point calculations.

The boxes on the grid overlap in terms of index numbers.  LSAT boxes cover a 5 point range and overlap by 2 points -- e.g., the next lower box on UCLA's grid covers index numbers 190-196. GPA boxes cover a 2.5 point range and overlap by 4 or 5 points -- e.g., the box to the right of our starting point covers index numbers 192-198.  So once I calculated each box, I combined the results and made a master chart.

Here's what you get:

Caveat:  I take sole responsibility for all the number crunching implied in this chart.  UCLA has nothing at all to do with the chart below. UCLA doesn't even know that this chart exists!  Is that clear?

White bars are applicants, yellow are admitted, orange are enrolled.  Number of apps runs up the left, and index numbers (from 220 down to 160) run across the bottom.

This is what the admissions officer sees, not the grid in the Official Guide.  So now let's imagine ourselves in  former Dean of Admissions Michael Rappaport's shoes.

• Index # 205-6  -- no problems -- accept, but we probably won't get that one.
• Index # 202-4 -- review fast -- this is the top of our class.  
• Index # 199 -- review pretty quick -- we want a bunch of these people.
• Index # 197 -- toss the weak ones; set the strong ones aside, see how we do with the higher numbers.
• Index # 195 -- reject most, hold a few really interesting ones for later review.

From about 198 to 204, early apps have an advantage -- a lot are being taken, but far from all.  Above 204, we'll accept them any time, and probably won't manage to enroll them.  Below 198, we're only taking a handful, we're their top school, and we'll get them even if we take them late.

So in a 7 index point range, early apps are favored.  How many apps are in that range?  By my calculations, 750 out of 4200.  That's about 18%.  For the rest of you, early doesn't matter.  

Now let's get back to our original questions -- why do admissions officers stress applying early?  Because of those 750, they're accepting about 400 and enrolling about 175 -- 2/3 of their class!  So they see that 2/3 of their class comes from the files they decide early.  I see that only 18% of applicants get an advantage by applying early.  

I hope that helps, because it was an awful lot of work!