Standard deviation of mean matriculated applicant GPA?

[empty p orbital]

So we've all looked at the MSAR stats of mean GPA/MCAT scores of accepted students at our various med schools of interest, but does anyone know where you can find data on the standard deviation for those numbers?

Say a school has an average accepted GPA of 3.8... is the SD like .1 where basically 95% of all their accepted students had between a 3.6 and 4.0? Or is it a bit larger, say an SD of like .3, where the spread is a lot wider and roughly a third of matriculants had a GPA LOWER than 3.5?

Imagine a student had a 3.3 cum GPA and a 3.2 science gpa (this is not me - I'm just being hypothetical here). Reading the posts on this forum, most would come away thinking the applicant has no shot at all unless there are some extreme extenuating circumstances. Is that presumption really correct? Or could that applicant potentially have a decent chance if he has an otherwise great application with very good MCAT scores, plenty of quality ECs, good LoRs, etc. and had a personality that was impressive and desirable by adcoms?

So speaking about standard deviation and GPA only, is an applicant like that getting accepted really an out-lier with only 3 or 4 applicants getting accepted with that gpa per school each year? Or is the deviation larger, with 20-25% of the accepted students coming with GPAs between 3.2-3.4?

---

Members don't see this ad.

 

[drizzt3117]

Generally the ranges are reasonably tight. I'd advise targeting your schools such that you aren't more than .2 or at most .3 off their ACCEPTED means, for the vast majority of your schools.

 

[loveoforganic]

0.1 sounds like a reasonable standard deviation to me, but just guessing. Would be a useful stat to know. I'm sure it's pretty negatively skewed though

 

[addo]

I came across a stat sheet that gave the SD, but i dont remember if it was aggregated stats or stats for a specific school. Anyways from what I remember the SD was about .15 in that specific case, which would seem reasonable, though not for me.
If the average GPA for accepted applicants is a 3.6, a .2 SD would be feasible.

 

[mmmcdowe]

If you had the time you could get a reasonable calculation by looking at the 10th and 90th percentiles of every school in the MSAR and averaging them based on the number of matriculated students per year. That is unless the MSAR already has those values available.

 

[RogueUnicorn]

if you look at the 10th and 90th percentile values given in the mcat for a given school, you can back-derive the standard deviation. the difference between these two values should be ~3.2 of the stdev

edit: beaten to the punch by mmmmmmmmcdowe

 

[Longshanks]

You should be more concerned with accepted applicant data, than matriculated. Matriculated data can be affected by many things, such as cost of attendance or "prestige factors" if applicants have more than one acceptance. Accepted applicant data will then be of more use for you targeting schools. MSAR reports accepted applicant data.

The 10th and 90th percentile information can also be of use. You'll notice there may be instances where schools have the same median for gpa/mcat but their spread can be vastly different. There's even some instances in MSAR where a school with a higher median actually has a greater variability than a school with a lower median. Just some food for thought.

 

[loveoforganic]

if you look at the 10th and 90th percentile values given in the mcat for a given school, you can back-derive the standard deviation. the difference between these two values should be ~3.2 of the stdev

edit: beaten to the punch by mmmmmmmmcdowe

My stats skills are rusty - would this only work with normal distribution?

 

[RogueUnicorn]

My stats skills are rusty - would this only work with normal distribution?

i think it's a decent enough assumption, though the data is surely skewed

 

[BigEast55]

i think it's a decent enough assumption, though the data is surely skewed

A reasonable estimate may be to do something like
((mean-10th percentile)+(90th percentile-mean))/(2*2)

This would average out what is certainly skewed towards the lower end for more schools as you cant get above a 4.0 and most schools are pretty close to that.

But for a school like Duke, it would give a st. dev. of 0.13 which is pretty reasonable I feel like I would imagine most schools are between 0.1 and 0.2

 

[ownage]

You can easily calculate the standard deviation from MSAR. Or have general idea of what the standard deviation is from looking at 10th to 90th percentile. You really dont need standard deviation if u have MSAR