AMCAS Statistics Refinement: MCAT vs GPA vs Acceptance

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.
Excelsius

Excelsius

Carpe Noctem
10+ Year Member
15+ Year Member
Joined
Apr 6, 2008
Messages
1,667
Reaction score
12
There is another thread floating around here concerning this Graph. Many people seem to to be confused about the data and what they are supposed to represent. Below I am providing graphs that separate out the numbers of URMs and non-URMs. As you will see yourself, in some cases the differences can be large, while in others they are negligible. The short answer to "Do I have a 20% chance of getting into medschool with below 20 MCAT and over 3.8 GPA" is no.

Enjoy:

Gr.jpg




Non-URMAcceptance.jpg


URMAcceptance.jpg


URMvsNONvsMCAT.jpg


URMvsNONvsGPA.jpg


MCATvsGPA.jpg


Note that the last graph finally puts the MCAT vs GPA question to rest. The choice of standardization here has been very generous: it was assumed that 2.0-2.19GPA=5-14MCAT, 3.8-4.0GPA=39-45MCAT. Considering that 20.79% of applicants have GPAs over 3.8 but only 1.47% have MCATs over 39, this graph would look much more favorable towards MCAT if I spent the time properly standardizing GPA vs MCAT correlation. I am attaching the Excel file in case some of you guys want to play with the data further.
Members don't see this ad.
 

Attachments

  • AMCAS Acceptance.xlsx
    74.5 KB · Views: 424
I'd be surprised if the people asking the stupid question to begin with would be able to interpret this information to reach a reasonable conclusion.
 
Members don't see this ad :)
Looks like the only real difference is from an MCAT of about 18ish to low 30s
 
I think the most interesting things about these graphs are

Most interesting:
In the limit of high MCAT scores any GPA over 3.4 is essentially the same

Also
I) How rapidly the marginal benefiet of getting over a 30 on the MCAT declines for most GPA groups
II) Except for people who have a 3.2-3.4 GPA who experience a steady rate of marginal benefiet with increasing MCAT scores
III) How much better a 3.5 is than a 3.3

Seeing these graphs made me feel alot better an engineer seeing as I will be in lower GPA higher MCAT regime, I just need to do everything i can to get to the 3.41 barrier, but even if i don't im not sunk seeing as of the people who are getting in with 3.2-3.4 they are probaly people in hard majors at prestigous universities.
 
If I have some free time im curious to get some regressions of those curves and get some graphs up of their derivatives and second derivatives. Id be curious to get a better representation of the "tipping point" areas and see the inflection points and the such. Just from eyeballing the graphs it seems like there could be some interesting trends in their derivatives.
 
surftheiop said:
If I have some free time im curious to get some regressions of those curves and get some graphs up of their derivatives and second derivatives. Id be curious to get a better representation of the "tipping point" areas and see the inflection points and the such. Just from eyeballing the graphs it seems like there could be some interesting trends in their derivatives.
Click to expand...
That's a nice idea, but I am not sure if it can be done in Excel. The only estimation of an equation is through the trendline, which only works for simple graphs. If you are able to do this, I'd be curious to know how you did it. I am thinking that the only way to do this would be to use a program like Maple. I have never used it to analyze graphs though and this will be pretty time consuming. While the equation would not be static from year to year, it should be a better success approximation than the LizzyM equation. Then again, given the amount of work required, it should just suffice to look up any particular number using one of the graphs. I think it will be a fun summer project to convert medical schools admissions into one big equation - the theory of everything.
 
Excelsius said:
That's a nice idea, but I am not sure if it can be done in Excel. The only estimation of an equation is through the trendline, which only works for simple graphs. If you are able to do this, I'd be curious to know how you did it. I am thinking that the only way to do this would be to use a program like Maple. I have never used it to analyze graphs though and this will be pretty time consuming. While the equation would not be static from year to year, it should be a better success approximation than the LizzyM equation. Then again, given the amount of work required, it should just suffice to look up any particular number using one of the graphs. I think it will be a fun summer project to convert medical schools admissions into one big equation - the theory of everything.
Click to expand...

You'd be better off using matlab which is much better with arrays (i.e. data sets). :) Maple is a program that's better suited for symbolic math.
 
Food said:
Great. Another annoying-ass poster who puts esoteric, cryptic messages with the hope that someone like him will respond in kind.
Click to expand...

He's talking about another data analysis program. Origin is often used instead of excel because it has better math capabilities, but isn't as crazy as matlab.
 
Excelsius said:
That's a nice idea, but I am not sure if it can be done in Excel. The only estimation of an equation is through the trendline, which only works for simple graphs. If you are able to do this, I'd be curious to know how you did it. I am thinking that the only way to do this would be to use a program like Maple. I have never used it to analyze graphs though and this will be pretty time consuming. While the equation would not be static from year to year, it should be a better success approximation than the LizzyM equation. Then again, given the amount of work required, it should just suffice to look up any particular number using one of the graphs. I think it will be a fun summer project to convert medical schools admissions into one big equation - the theory of everything.
Click to expand...

Boo excel had my hopes up, I saw that it had a polynomial choice for trendline and I was just going to brute force pseudo- "taylor series" the thing by setting the order of the polynomial to like 25, but sadly the highest order it would let me do is 6 :(
6th degree polynomial still actually fit data reasonably well but not good enough for getting meaningful data from derivatives.

Its been awhile since ive done it, but several of those curves could probably be decent for fourier approximation (they look like part of a sine curve that has been shifted up off the axis) but I don't rember what preconditions are needed for that?
 
Esoteric is a relative term. "Origin" is not esoteric to me, and if it is for you, I truly feel sorry for you :(:(:(.
 
Let's keep in mind that this is a pre-medical forum. As such, it will be a good idea (and courteous) to use complete sentences when describing very major specific items, especially if their names can be confused with something else. With a physics major, I know about a lot of software applications like Matlab, Mathematica, Scientific Workplace, etc, but I hadn't heard of Origin before. Ambiguity can also be intentional at times, for whatever reason.
 
Anyway, forget it. My apologies to starfish. I in fact quite enjoy your 1 word responses to posts, they make me giddy
 
Im kinda surprised this thread never went anywhere. very useful work excelsius.
 
You must log in or register to reply here.
Next unread thread

Similar threads

A
Consequences with AMCAS for violating an AACOMAS acceptance
Replies
23
Views
1K
RunnerDad
R
F
Question about statistics requirement
Replies
1
Views
315
whichbreath
W
hi.985
Would this count towards my BCPM gpa for AMCAS/TMDSAS?
Replies
3
Views
536
Holland
Holland
hi.985
Would this class count towards my science GPA (BCPM) for AMCAS & TMDSAS?
Replies
4
Views
653
Mr.Smile12
Mr.Smile12
j_diggity
Different addresses on TMDSAS vs. AMCAS?
Replies
2
Views
519
wysdoc
wysdoc
Top