EDIT (kryzak): This is a very cool analysis by rhyme and pelihu with a lot of good information on how the GMAT relates to UE/E schools' acceptance rates and yield. Doesn't mean you can slack off with your essays and recommendations, but it may help you sleep better at night if you have a 700+ GMAT score.
Have I reverse engineered Kellogg's accept rate by GMAT score? NO, but I may have stumled on something. SKIP TO THE SECTION CALLED "COOLNESS" to skip over all my crazy attempts to back into an answer. Someone good at math, please jump to that section and tell me if im nuts.
I'll walk you through my entire logic to show you how I got here.
Take the percentages given in the Kellogg PDF to figure out total applicants per GMAT grouping. Take that and use their "enrolled" percentage to give you total enrolled per GMAT grouping. The sum of these figures SHOULD equal the total enrolled (652). Instead, it equals another number, 1350 or so. It just so happens thats Kellogg's yield percentage from 2004, applied to this number, gives you roughly the correct number of individuals that you would expect to see enrolled. The implication: The enrolled percentages used in the table are in fact, the accept percentages by gmat score, not enrollment percentages by gmat score. THIS IS WRONG BECAUSE, IN ALL LIKELYHOOD, THE PERCENT ENROLLED IS NOT BROKEN OUT BY GMAT, BUT RATHER REFLECTS A TOTAL AVERAGE, BUT IT LEADS ME TO AN INTERESTING ALTERNATE CONCLUSION.
Lets get started:
https://www.kellogg.northwestern.edu/adm ... res_05.pdf
Start with this document.
Total Applicants 4449 (provided in the document)
Total Enrolled 652 (provided in the document)
Enrollment rate 14.7% (calculated 652 / 4449 --- not an accept rate, but an enrollment rate)
Now lets look at the GMAT table in that PDF.
Of applicants who scored 640 or less, 9% enrolled, and they made up 20% of the applicants. Of those who scored, 650 to 690, 25% enrolled, of the total applicants they made up 28%. For 700-740, 49% enrolled, 41% of total apps had this score. Same kind of stuff for 750-800.
Based on the total applicants # (4449), and the total applicants %'s we can back into the number of applicants per score group.
< 640: 890 applicants
650-690: 1246 applicants
700-740: 1824 applicants
750-800: 489 applicants
I get this by just taking the total (4449) and multiplying the percentage in that table. I.e. 4449*20% = 890 (the numbers for 640). As expected, this sums to 4449 (obviously since im just reversing the math).
Now lets do the same thing for enrolled #. Lets take the < 640 numbers first as an example.
We know 9% enrolled.
We know 20% of applicants had a score < 640.
We just figured out that there were 890 applicants (4449*20%) with a score < 640.
Thus, total enrolled students with a GMAT score under 640 should be about 80 (890*9%). THIS IS WHY THIS ANALYSIS IS PROBABLY WRONG. ODDS ARE, THIS 9% REPRESENTS 9% OF 652, NOT OF 890. BUT FOR THE SAKE OF ARGUMENT, ILL CONTINUE.
Do the same for the others. You get:
< 640 - Total enrolled: 80
650-690 - Total enrolled: 312
700-740 - Total enrolled: 894
750-800 - Total enrolled: 78
So now we have an idea of the total number of students enrolled. But wait, this number comes to 1364 (add them all up), not to 652 - the total enrolled on the main page of the PDF!. Shouldn't this add up to 652?
Now look here:
https://www.businessweek.com/@@ZuBnf4QQP ... ellogg.htm
Says its yield is 57% , 61%, 64% over 2004, 2003, 2002.
Now things get confusing, but work with me here.
So what if you take 57% of 1364? What do you get?
You get 777.
Whats 777 / 4449 = ? It's the net enrollment rate right?
Whats this percentage? 17%.
Whats the net enrollment rate we calculated BASED ON THE NUMBERS IN THE PDF? 652 divided by 4449 is 14.7%.
Well, 17% is mightly close to 14.7% - especially when you consider that we are only estimating 2005 based on 2004 numbers - the error here - less than 3%, makes sense given this fact.
So what does this imply?
This implies 1364 is the number of accepted applicants, not the number of enrolled!
The logical implication is that the percentages in the table are NOT ENROLLED PERCENTAGES, BUT RATHER, ACCEPT PERCENTAGES. This cant be
right?
You might ask - well then, why isn't it exactly 14.7%? We are estimating based on the yield of 57% (cause we dont have 2005 yield percentages on businessweek, so im using 2004s).
Take it another way - 652 is what percent of 1364? About 48% right? Thats a reasonable yield for 2005 right?
One other way to look at it - the accept rate in 2004 was 23%. Using this number as a proxy for 2005, we would expect 1023 accepts (4449*23%).
If you believe that the % in the enrolled column is in fact the accept %, then we come to 1364 accepts, or a 31% acceptance rate - possible if you consider the lower yields!!! Your accept rate will correlate in exactly this manner with the yield. That is, 2004 had 23% accept and 57% yield. In 2005, if KGSM started experiencing a lower yield (48% if you do the math), it would follow that their accept rate would need to be slightly higher to compensate. The numbers make sense in this regard.
BUT THIS JUST ISNT RIGHT, IT JUST CANT BE. SO IGNORE IT.
NOW FOR COOLNESS.
But here's something else MUCH MORE INTERESTING THAT FOLLOWS.
What if we buy that enrolled % is just that - enrolled %?
Then, if there are 652 enrolled, and those who got a 640 make up 9%, then we'd expect 59 students from the < 640 group enrolled. Knowing kellogs YIELD (NOT selectivity) hovers around 57%, that means that we would expect for those under 640, that 103 offers were made, and 59 attended.
Do the same for the others:
640: 103 accepts
650-690: 286 accepts
700-740: 560 accepts
750-800: 183 accepts
Total accepts ~ 1132 (which, if you take note, IS ROUGHLY 25%, WHICH IS EXACTLY WHAT YOUD EXPECT!!)
So now we know this...
Out of the 890 applicants who applied with less than 640, 103 got accepted. Hence, if you fall into this bucket, your odds of acceptance are 11%.
Do this for the other numbers:
640: 103/890 = 12%
650-690: 286/1246 = 23%
700-740: 560/1824 = 31%
750-800: 183/489 = 37%
Voila!
Anticipated accept odds by gmat level!
Whadda ya think? Math geeks? Am I nuts?
Now here's something even more interest.... (coming soon)