Serious differences in MSAR vs. School Reported Stats.....

[Detective SnowBucket]

Hi all, I hear you like tables and data so I've compiled a list of schools I'm looking at for myself and went through and gathered up all their relevant data through the MSAR then went and double checked with the school's own sites.
I have looked into other threads on SDN and the general consensus is trust the MSAR because the school site may not be updated. I found updated (as in 2017 or sometimes 2016 cycle) for nearly all the schools and put a link to their pages. There are just some serious differences with GPA's mostly and I'm inclined to believe the updated school's page.

The color coding is green = safety, light blue = target & dark blue = reach (...according to MSAR stats). Crossed out means I'd have no chance according to MSAR stats. Didn't bother with OOS % for private schools.
@Goro , I know you have some thoughts on the MSAR and I've put some effort into this so I'd appreciate some help, Thanks

---

Members don't see this ad.

 

[efle]

Can you give me an example of a school reporting a value very different from the MSAR for a median for the same cohort (either accepted vs accepted, or matriculated vs matriculated)?

Edit: It is not possible for the MSAR to ever be wrong on these values, btw. It's the same group that runs AMCAS. They are working from the raw data, they have all the correct GPA and MCAT values. If a school reports a different number, in every case I've ever seen it looks like it's because it is apples vs oranges (for example, enrolled average vs admitted median). As another example, the Einstein page at the top is for the class of 2017...meaning the people that graduated six months ago. That's a 4 year old value. The duke link a little below that is for MSTP, not for regular MD...

 

[Goro]

Hi all, I hear you like tables and data so I've compiled a list of schools I'm looking at for myself and went through and gathered up all their relevant data through the MSAR then went and double checked with the school's own sites.
I have looked into other threads on SDN and the general consensus is trust the MSAR because the school site may not be updated. I found updated (as in 2017 or sometimes 2016 cycle) for nearly all the schools and put a link to their pages. There are just some serious differences with GPA's mostly and I'm inclined to believe the updated school's page.

The color coding is green = safety, light blue = target & dark blue = reach (...according to MSAR stats). Crossed out means I'd have no chance according to MSAR stats. Didn't bother with OOS % for private schools.
@Goro , I know you have some thoughts on the MSAR and I've put some effort into this so I'd appreciate some help, Thanks

If the website is giving you the old MCAT score, and MSAR is giving you the new MCAT score for medians, what does that tell you of the veracity of the data?

the numbers are almost the same. I mean mean, differing by 0.1 for GPAs? Not anything to write home about.

 

[Detective SnowBucket]

Yea, looks like you're right about those two, my mistake. Nearly none of the school reported pages give the cohort but regardless of cohort (be it one or the other, or even the mean value of the two) shouldn't the values still be close. The best example is Georgetown, reports a 3.55 sGPA but MSAR has 3.66 and 3.74. Also, UCLA reports a 517 average but MSAR shows 508 and 505. Michigan reports 3.71 sGPA and 3.77 cGPA but MSAR has its GPS's between 3.8 and 3.86. These values aren't even close, can you explain how cohort would shift values that much?

 

[Biological]

Yea, looks like you're right about those two, my mistake. Nearly none of the school reported pages give the cohort but regardless of cohort (be it one or the other, or even the mean value of the two) shouldn't the values still be close. The best example is Georgetown, reports a 3.55 sGPA but MSAR has 3.66 and 3.74. Also, UCLA reports a 517 average but MSAR shows 508 and 505. Michigan reports 3.71 sGPA and 3.77 cGPA but MSAR has its GPS's between 3.8 and 3.86. These values aren't even close, can you explain how cohort would shift values that much?

Remember that schools may also have different criteria for how they determine/report MCAT and GPA. For example, for MCAT score the MSAR uses the most recent MCAT score in their calculations - a particular school might use the highest not necessarily the most recent. For GPA, a number of schools recalculate GPA according to their own specific set of criteria.

 

[Goro]

Yea, looks like you're right about those two, my mistake. Nearly none of the school reported pages give the cohort but regardless of cohort (be it one or the other, or even the mean value of the two) shouldn't the values still be close. The best example is Georgetown, reports a 3.55 sGPA but MSAR has 3.66 and 3.74. Also, UCLA reports a 517 average but MSAR shows 508 and 505. Michigan reports 3.71 sGPA and 3.77 cGPA but MSAR has its GPS's between 3.8 and 3.86. These values aren't even close, can you explain how cohort would shift values that much?

Haven't a clue. Suggest asking the people who maintain the websites at those schools.

 

[efle]

Yea, looks like you're right about those two, my mistake. Nearly none of the school reported pages give the cohort but regardless of cohort (be it one or the other, or even the mean value of the two) shouldn't the values still be close. The best example is Georgetown, reports a 3.55 sGPA but MSAR has 3.66 and 3.74. Also, UCLA reports a 517 average but MSAR shows 508 and 505. Michigan reports 3.71 sGPA and 3.77 cGPA but MSAR has its GPS's between 3.8 and 3.86. These values aren't even close, can you explain how cohort would shift values that much?

Georgetown has one of the big SMP programs, and admit several from that every year. So I absolutely would expect their mean to be ~0.2 below their median. You only need a few people coming in from the far left side of the curve to start significantly skewing the average down, and if anyone is going to get into med school with a low GPA, it's the SMP people.

For UCLA, that is just straight up not a valid source! Medical school admission.com ?? Absolutely trust the MSAR more

Michigan again I'd say -0.1 is a very reasonable skew to see on the mean compared to the median. Most med school classes are going to have a much longer left tail. In fact we can see this is true in the MSAR - as an example look at Michigan. Their 90th is only +0.15 over the median, while the 10th is -0.25 below it. That's gonna be a left skewed distribution.

 

[gyngyn]

Trust MSAR.

 

[boogiecousins94]

the only thing that may be different (from what I've noticed) between school sites and the MSAR is the number of total applications to each school.

MSAR lists the total applications regardless of secondary completion. Some school sites on their class profile list "completed applications" so the number is actually lower. In some cases its ~1000 less (Tufts is one of them, 11,029 initial apps, 9775 completed 2021 Class Profile | Tufts University School of Medicine )

 

[Detective SnowBucket]

Haven't a clue. Suggest asking the people who maintain the websites at those schools.

It sounds like you're saying a 0.1 difference isn't all that big a deal; do you think it wise to apply to a school with a 3.8 average when you are at a 3.7 - just for the sake of example - ?

 

[DBC03]

Three big differences between MSAR and school sites - MSAR is from the class that matriculated two years ago, focuses on the median (not mean), and includes every application. Schools may be giving you data on the most recent class, the mean and not median, and they may only include data from those who completed secondaries or those matriculated, etc....

Also, a majority of the MCAT scores from two years ago were from the old MCAT, so the 2015 MCAT numbers represents a small fraction of those accepted (in the MSAR). Newer numbers from the schools themselves may be more realistic. I expect the numbers out in two years will be the best representation of what the schools are accepting overall.


Sent from my iPhone using SDN mobile

 

[efle]

Applying with a 3.7/516 to a school with median 3.8/516 is totally appropriate !

 

[skk_]

Yea, looks like you're right about those two, my mistake. Nearly none of the school reported pages give the cohort but regardless of cohort (be it one or the other, or even the mean value of the two) shouldn't the values still be close. The best example is Georgetown, reports a 3.55 sGPA but MSAR has 3.66 and 3.74. Also, UCLA reports a 517 average but MSAR shows 508 and 505. Michigan reports 3.71 sGPA and 3.77 cGPA but MSAR has its GPS's between 3.8 and 3.86. These values aren't even close, can you explain how cohort would shift values that much?

UCLA's own admissions site flat out tells you to use MSAR directly to get their admission statistics and doesn't post it themselves, so I have no idea where you got your stats for UCLA from.

Pitt (in your xls) has a guaranteed admissions program for high school seniors that (I believe) does not require a full AMCAS application to even be filled out - this likely accounts for the disparity of their incoming class GPA/MCAT on their site (which includes these students) and MSAR (which likely includes people who applied normally only). I imagine that this might be the case for several other schools as well.

 

[Goro]

It sounds like you're saying a 0.1 difference isn't all that big a deal; do you think it wise to apply to a school with a 3.8 average when you are at a 3.7 - just for the sake of example - ?

Yes.

You do understand what medians, right?

Even a 3.6 can get you into Harvard or Stanford.

 

[Detective SnowBucket]

I understand what median means, I'm just hesitant to rely on it as the best representation of a data set

 

[efle]

I understand what median means, I'm just hesitant to rely on it as the best representation of a data set

Nothing really substitutes for a full box-and-whisker or curve, but if you have to decide between using a mean and a median for a dataset you know is not symmetric, the median is always the better option. More resistant to skew!