composite MCAT score percentile

[AegisZero]

Anyone know where you can find out your composite MCAT score percentile from August 2002? Or if you have a vague idea of the general percentiles, anything would be appreciated.

Thanks

---

Members don't see this ad.

 

[Bevo]

dont wanna know.

 

[JScrusader]

Here is the link for a whole bunch of data including the August MCAT:

http://www.aamc.org/students/mcat/examineedata/pubs.htm

If you assume that the composite mean is the same as adding all of the section mean scores with the standard deviation as 2.3 and a normal distribution, you can use excel to find percentiles.

I am not sure how accurate it is.

 

[JScrusader]

This was calculated using the combined April/August 2002 data.

Mean Std Composite Percentile
24 2.4 20 0.0477903
21 0.1056498
22 0.2023283
23 0.3384612
24 0.5000000
25 0.6615388
26 0.7976717
27 0.8943502
28 0.9522097
29 0.9813896
30 0.9937903
31 0.9982310
32 0.9995709
33 0.9999116
34 0.9999845
35 0.9999977
36 0.9999997
37 1.0000000
38 1.0000000
39 1.0000000
40 1.0000000

 

[CaNEM]

Originally posted by JScrusader
This was calculated using the combined April/August 2002 data.

Mean Std Composite Percentile
24 2.4 20 0.0477903
21 0.1056498
22 0.2023283
23 0.3384612
24 0.5000000
25 0.6615388
26 0.7976717
27 0.8943502
28 0.9522097
29 0.9813896
30 0.9937903
31 0.9982310
32 0.9995709
33 0.9999116
34 0.9999845
35 0.9999977
36 0.9999997
37 1.0000000
38 1.0000000
39 1.0000000
40 1.0000000

These numbers are hard to believe. First of all, the average score for all takers is typically around a 24, and for those accepted to med school, around a 30 (across all US schoools). A percentile score of 0.9937903 for a score of thirty would imply that the average student accepted to school is in the top one percent of test takers! Also, the average score in both the April and August tests was not EXACTLY a 24, so claiming a percentile as perfect as 0.5000000 is ridiculous. Finally, these numbers imply that only about 3 out of 10,000,000 people score a 36 or higher (since the percentile listed for 36 is 0.9999997)! That's impossible, because the number of people who take the test is only in the tens of thousands, NOT the millions, and there is certainly more than one person who got a 36 or higher out of these. Is this fuzzy math, or am I missing something?

 

[JScrusader]

This is what excel spit out. I think it is accurate. There was a very large population. 57,571 people took the test in april or aug. Adding the individual mean scores equals 24 exactly. You can also look at z scores which tells you how many standard deviations above the mean a score is. For example, a 30 would be (30-24)/2.37 =2.53 std devs above the mean. From my stats book, the area below a z score of 2.53 would be .9943 which is very close to the excel number.

 

[ttac]

You actually can't do it that way...
The SD for _each_test_ may be 2.5, but that doesn't mean that if you got 2.5 points above the mean of 24, then you are one SD above the mean, because you are getting contributions from each test.

Theoretically, if you get 2.5 points (or whatever the SD is) above the mean on each test, you should get around a 31, which would be one SD above the mean...

Of course, people that do well on one section are more likely to do well on another section, so the scores are not independent, and you can't really estimate percentages from the SD of each test.

I remember reading somewhere (now you KNOW this is going to be an ironclad, irrefutable point ) that there were something like 200+ people out of the 40,000 people that took the MCAT last year that got >=40... Correct me if I am wrong.

-ttac

 

[JScrusader]

The problem is whether to add all of the std dev together or to take an average of them. adding them would give 7.1 and would bring the z score for a 30 to a .845. the area below .845 is about .8 or the 80th percentile. a 36 would be the 95.5th percentile.

 

[ttac]

I think there really isn't any way to calculate composite percentages from the distribution on each subsection, b/c although it seems intuitive (i.e. I'd bet good money on it ) that people who score well on one section would score well on another, you don't know how strong that correlation is, so what you really need to do is email the MCAT people and ask them to tell you how many people scored 30, 31, 32, 33, etc...

Volunteers?

-ttac

p.s. 36 as 95% percentile seems about right...

 

[CaNEM]

JScrusader, your math is flawed, and it's impossible to make the conclusions you did. To prove this to yourself, read my prior post. According to your numbers, NO ONE out of the 50,000 would get >36 and LESS THAN 1% of people would get >30. You know that is silly.

 

[AegisZero]

what percentile would you guess a 38 or a 39 is at?

 

[CaNEM]

Originally posted by AegisZero
what percentile would you guess a 38 or a 39 is at?

About 250 people out of the 50,000 test takers score a 39 or above. This is the 99.5th percentile (0.995).

 

[JScrusader]

38-24/7.1=1.97 area below 1.97=.9756
39-24/7.1=2.11 area below 2.11=.9826

I dont't think the math is flawed. There must be some type of inaccuracy or statistical impossibility in determining the composite percentile or they would give it to us. As far as adding the std dev. together, i think it makes sense, thus this method should give a rough estimate. But those who are still in school, please go ask you statistics professors how to do this.

 

[JScrusader]

where did you get 250 people from?

 

[AegisZero]

the new MSAR has a graph that shows your the number of applicants with a certain MCAT score (even the composite). The problem is its very hard to read. I have made the following conservative (meaning I had more people in a category than less) estimates for these scores out of 35,000 takers

38: 900 students at or above
39: 500 students at or above

If anyone is interested in me tabulating other scores I am willing to make the attempt, or you can find this in the MCAT book

 

[AegisZero]

I would like to add that the MCAT score percentiles I am providing are for actual applicants. That skews the percentiles downward since we can guess that the lower MCAT scores don't even apply to med school.

 

[CaNEM]

Originally posted by JScrusader
where did you get 250 people from?

I estimated, from memory. It's actually closer to 400, based on these numbers someone posted in the preallo forum.

42 / 26 / 26
41 / 53 / 53
40 / 105 / 105
39 / 211 / 211
38 / 395 / 316
37 / 579 / 474
36 / 711 / 579
35 / 974 / 789
34 / 1368 / 1105
33 / 1684 / 1237
32 / 2026 / 1421
31 / 2395 / 1658
30 / 2474 / 1632

 

[AegisZero]

what is the difference between the value given in the left column and the value given in the right column? where were these stats taken from (like which post)? thanks!

 

[CaNEM]

Originally posted by AegisZero
what is the difference between the value given in the left column and the value given in the right column? where were these stats taken from (like which post)? thanks!

I believe they are estimates of people who apply and people who get accepted. The original post was called something like "the 40+ myth."

 

[Negrodamus]

WTF is up with this thread necromancy? I was still in middle school when this thread was made.

 

[NYR56]

WTF is up with this thread necromancy? I was still in middle school when this thread was made.

lmao. I think he's just a spammer, that link is certainly not helpful and he posted the same exact thing in another thread.

 

[john1096]

Here is the link for a whole bunch of data including the August mcat:

http://www.aamc.org/students/mcat/examineedata/pubs.htm

If you assume that the composite mean is the same as adding all of the section mean scores with the standard deviation as 2.3 and a normal distribution, you can use excel to find percentiles.

I am not sure how accurate it is..

this above aamc.org link is not working..
for latest 2010 mcat results there is an information on the home page of aamc.org website..