Efle's MCAT 2015 to Old MCAT Percentile Comparison/Conversion Tables

[efle]

Updated 5/3/2018

Good evening fellow premeds, wise med students/docs/adcoms, and Goro.

I put together an excel file for easy comparison of the new MCAT 2015 and Old MCAT percentiles, both composite and section scores. It is built from AAMC's official distributions: the Final Percentiles for the MCAT 2012-2014 and the New Percentile Ranks for May 2018-April 2019.

Here are the tables for quick reference:

Let me know any suggestions or mistakes.

---

Members don't see this ad.

 

[HCHopeful]

THIS IS PURE GOLD. NOW PEOPLE WILL STOP WHINING.

Lol. Not really. They always whine.

 

[Lawpy]

It's probably a plus that a 13-15 on old VR is equal to only 132 on CARS lol. Kinda gives some room for error, I'd hope, addressing @ElCapone and @gettheleadout claims on how subjective and error-sensitive verbal is.

 

[efle]

It's probably a plus that a 13-15 on old VR is equal to only 132 on CARS lol. Kinda gives some room for error, I'd hope, addressing @ElCapone and @gettheleadout claims on how subjective and error-sensitive verbal is.

It's extremely oversensitive, they used to report it as "13-15"

 

[gyngyn]

Good evening fellow premeds, wise med students/docs/adcoms, and Goro.

I put together an excel file for easy comparison of MCAT 2015 and Old MCAT percentiles, both composite and section scores, available here. It is based on the Final Percentiles for the MCAT 2012-2014 and the April and May MCAT 2015 Percentiles.

There are two types, Compressed and Expanded.

Compressed lists only the percentiles for which there are corresponding scores on one of the two tests. This is the easiest way to see what old scores a new score lies betwixt (or equal to). Example Verbal/Cars Compressed:

Expanded lists all percentile values 0-100, with the corresponding scores from each test spread throughout it. This is not as quick to read, but lets you see visually how close a new score lies to the old scores above and below it. Example segment from the Composite Expanded (again left to right is Percentile, New score, Old score):

In case excel doesn't work for y'all, all of the expanded can be found as images here, and the compressed here.

Let me know any suggestions or mistakes.

Well it looks like you and I have been doing exactly the same thing today.
With remarkably similar results, I might add.
I additionally assigned screeners by score, though!

 

[Lawpy]

132 is the highest possible.

It's extremely oversensitive, they used to report it as "13-15"

You need to get several more questions wrong to drop from 132 (13-15 equivalent) to a 131 (~12 equivalent). And only 1-2 questions wrong will still assure a 132.

 

[deleted393595]

It's probably a plus that a 13-15 on old VR is equal to only 132 on CARS lol. Kinda gives some room for error, I'd hope, addressing @ElCapone and @gettheleadout claims on how subjective and error-sensitive verbal is.

THEY FINALLY FIXED THAT SCORING ISSUE YAYYYYYYYYYYYY!

 

[samac]

I lol'd at "and Goro"

 

[Lawpy]

THEY FINALLY FIXED THAT SCORING ISSUE YAYYYYYYYYYYYY!

Now if they will focus on removing the damn subjectivity in analyzing the passages...





... oh wait, that defeats the point of verbal! But hey, first step always helps

 

[efle]

I additionally assigned screeners by score, though!

Which were...?

Anyways, awesome to hear it'll be a similar tool to what adcoms might put together. Tag me if you share yours as well

 

[deleted393595]

Well it looks like you and I have been doing exactly the same thing today.
With remarkably similar results, I might add.
I additionally assigned screeners by score, though!

Proof that @efle would be a really good and capable student to have at your school *wink* *wink*

 

[efle]

Proof that @efle would be a really good and capable student to have at your school *wink* *wink*

I am a Cali resident...

 

[gyngyn]

Proof that @efle would be a really good and capable student to have at your school *wink* *wink*

An excellent administrator, perhaps...

 

[gyngyn]

I am a Cali resident...

I guess I'll see your application, then.

 

[gyngyn]

Which were...?

Anyways, awesome to hear it'll be a similar tool to what adcoms might put together. Tag me if you share yours as well

It's essentially the same.
The only additional work was the designation of strata to be evaluated by single examiners.
That and working on the software changes for data sort and reporting.
Rough day at the office. And the phone was ringing off the hook with applicants.
All this and @rachiie01!

 

[lantx]

I didn't think MD schools really screened out anything for MCAT. I mean perhaps a 27 or lower. Would you be able to estimate the cut offs for your school or other schools?

 

[HCHopeful]

I didn't think MD schools really screened out anything for MCAT. I mean perhaps a 27 or lower. Would you be able to estimate the cut offs for your school or other schools?

There are always cutoffs for specific schools. The MSAR is really your only buddy when it comes to this info.

 

[ComputerGuy365]

Can a mod please sticky this?

 

[gyngyn]

I didn't think MD schools really screened out anything for MCAT. I mean perhaps a 27 or lower. Would you be able to estimate the cut offs for your school or other schools?

A screener is not the same as a screen.
A screener is a human who reads and sorts applications.
A screen is a pre-set numerical cut-off.

 

[lantx]

A screener is not the same as a screen.
A screener is a human who reads and sorts applications.
A screen is a pre-set numerical cut-off.

Interesting. Screeners sort by MCAT score?

 

[gyngyn]

Interesting. Screeners sort by MCAT score?

Nope.
But screeners might be assigned statistically identical scores in order to control for the effect of the MCAT on their perception of the candidate's strengths.

 

[deleted393595]

A screener is not the same as a screen.
A screener is a human who reads and sorts applications.
A screen is a pre-set numerical cut-off.

So the process goes like:

1) Screen for GPA and MCAT
2) Screener reads PS and sorts for a certain MCAT range
3) Screener reads activities for each sorted pile and sorts even further?

 

[gyngyn]

So the process goes like:

1) Screen for GPA and MCAT
2) Screener reads PS and sorts for a certain MCAT range
3) Screener reads activities for each sorted pile and sorts even further?

That's one way to to it.

 

[deleted393595]

But screeners might be assigned statistically identical scores in order to control for the effect of the MCAT on their perception of the candidate's strengths.

That sounds scary if an applicant falls within a higher MCAT range since they're competing against (presumably) other strong applicants.

Unless it's more of a "weed-out" process to eliminate people with glaring weaknesses?

 

[gyngyn]

That sounds scary if an applicant falls within a higher MCAT range since they're competing against (presumably) other strong applicants.

Unless it's more of a "weed-out" process to eliminate people with glaring weaknesses?

Doesn't it seem reasonable to compare people with similar MCAT scores to each other in order to identify the "best" ones? This controls for the inevitable influence the score might have on the evaluator, thus achieving the elusive "holistic" evaluation .

 

[mw18]

That sounds scary if an applicant falls within a higher MCAT range since they're competing against (presumably) other strong applicants.

Unless it's more of a "weed-out" process to eliminate people with glaring weaknesses?

I would assume it is staggered by number of slots available and the more "high-yield" matriculant ranges get the most spots, and the less yielding getting fewer.

 

[lantx]

Doesn't it seem reasonable to compare people with similar MCAT scores to each other in order to identify the "best" ones? This controls for the inevitable influence the score might have on the evaluator, thus achieving the elusive "holistic" evaluation .

No! Because if one achieves a 99th+ percentile MCAT after messing up in school their GPA is lower. 99th percentile MCAT scorers are typically good students but there are others who haven't done as well in school.

 

[lantx]

I would assume it is staggered by number of slots available and the more "high-yield" matriculant ranges get the most spots, and the less yielding getting fewer.

This is tough for a discordant GPA/MCAT person.

 

[gyngyn]

No! Because if one achieves a 99th+ percentile MCAT after messing up in school their GPA is lower. 99th percentile MCAT scorers are typically good students but there are others who haven't done as well in school.

MCAT/gpa disparities are among the most challenging to assess.
It does raise the specter of some serious problems.
Your whole application will still be read.

 

[HCHopeful]

This is tough for a discordant GPA/MCAT person.

That's why it's best not to be "that person."

 

[lantx]

That's why it's best not to be "that person."

Too late for me. I don't even know what to do. Will low tier schools reject me because they think someone with my MCAT is likely to not matriculate? Even with my MCAT, top tier schools will reject me due to my GPA.

 

[HCHopeful]

Too late for me. I don't even know what to do. Will low tier schools reject me because they think someone with my MCAT is likely to not matriculate? Even with my MCAT, top tier schools will reject me due to my GPA.

It's been done before. Apply broadly and wisely. You'll be fine.

 

[Banco]

The distribution at the extreme ends is still pretty similar to the old MCAT. a 91% = 33 and a 97% = 36, while on the new one 91% = 514 and 97% = 518. However it is good to see a more spread distribution in the middle, whereas it was much more compact for the old one.

 

[Goro]

Kudos to you for the hard work, but your lead in line reminds me of the old joke:

"Ladies, Gentlemen, Efle..."

😉

Good evening fellow premeds, wise med students/docs/adcoms, and Goro.

 

[mw18]

This is tough for a discordant GPA/MCAT person.

I assumed mechanisms like the LizzyM score were used for discordant situations.

 

[gyngyn]

I assumed mechanisms like the LizzyM score were used for discordant situations.

This is an instance in which it doesn't hold up as well.
A big disparity has a very different differential diagnosis compared to more concordant sets of scores.

 

[mw18]

This is an instance in which it doesn't hold up as well.
A big disparity has a very different differential diagnosis compared to more concordant sets of scores.

Okay, makes sense. I just kind of thought that discordant ones would get put in a separate pile with, say, other 74 LizzyM's that have disparities. I appreciate all your insights. I've learned a lot. I'm already done applying, but it helps me make sense of what seems a very random process.

 

[efle]

This is an instance in which it doesn't hold up as well.
A big disparity has a very different differential diagnosis compared to more concordant sets of scores.

I imagine it also matters a lot which of the two is the discordantly high one. A 4.00 won't save your 27 the way a 37 can save a 3.00 etc

Regarding your earlier mention of strata in which readers see nothing but similar MCAT applicants...

how mediocre can you get with, say, a 37+, 3.9+ before your interview slot starts going to 30/3.5 that has better ECs? Is there some kind of numerical weight you give ECs so you can decide whether that LizzyM 70 deserves it over A LizzyM 80, since it's no longer at the discretion of an app reader that sees the whole range?

 

[gyngyn]

I imagine it also matters a lot which of the two is the discordantly high one. A 4.00 won't save your 27 the way a 37 can save a 3.00 etc

Regarding your earlier mention of strata in which readers see nothing but similar MCAT applicants...

how mediocre can you get with, say, a 37+, 3.9+ before your interview slot starts going to 30/3.5 that has better ECs? Is there some kind of numerical weight you give ECs so you can decide whether that LizzyM 70 deserves it over A LizzyM 80, since it's no longer at the discretion of an app reader that sees the whole range?

An executive decision-maker will calculate the number needed to interview from each of the sorted strata. There are no head to head comparisons of individuals.

 

[Dr. Retractor]

An executive decision-maker will calculate the number needed to interview from each of the sorted strata. There are no head to head comparisons of individuals.

Are applications stratified based only on numbers? Or are ECs taken into account there too?

 

[gyngyn]

Are applications stratified based only on numbers? Or are ECs taken into account there too?

We rely on the readers to stratify based on non-numerical information. That's why one might choose to give readers files from a single MCAT percentile to evaluate. Depending on the sophistication of the software used by a particular school the exact method of screening will necessarily vary.

 

[efle]

Depending on the sophistication of the software used by a particular school the exact method of screening will necessarily vary.

And do most schools use these sorts of complex algorithms and statistical analysis? I had this image in my head of some grumpy old dude with a black cat who reads through a random pile of apps and chooses his favorites to recommend for interview...did not realize there was sorting into stats strata and y̶i̶e̶l̶d̶ ̶p̶r̶o̶t̶e̶c̶t̶i̶o̶n̶ resource management via math model

 

[gannicus89]

Hmmm. I understand how having screeners/readers examine apps of students who have similar gpas/mcats might bring about a more holistic review...but what about the possibility of screeners' biases towards specific ECs or experiences? Basically... where is the inter-rater reliability? Shouldn't more than one screener be assigned to a particular gpa/MCAT bin to ensure a true "holistic" review? @gyngyn

 

[Grace184]

Wow gyngyn is getting grilled! Just give them the system, or you'll never escape this thread!

 

[Affiche]

@gyngyn it's a slow season at work for me 🙂

I have a similar but slightly different chart. The difference is that I don't rank new MCAT scores to be above or below old MCAT scores, but equal to. Example: 30= 508, 509, not 508< 30< 509

The reason is that with the wider ranges of the old MCAT, you can't rank scaled scores between tests with varied percentiles. With the wide range of a 30, most 508s are better than a 30, but some 30s are still better than a 508. You're just looking at the ceiling of that range, and that's inaccurate. If a score, like a 508, falls within the range of another score, such as 30, they are equal. It's the same process of converting for the old MCAT: 75th percentile fell in the range of a 30, so it's equal to all other 30s.

The ranking system in this thread fails to factor in that some 30s were near the 73rd/74th percentile, which is lower than a 508.

 

[Affiche]

I've finally been tagged by an adcom....life goal= complete

 

[Grace184]

I've finally been tagged by an adcom....life goal= complete

Just as long as you don't start a thread dedicated to them complete with a poem and expressing your second life goal of marrying them and spending the rest of your life with them...that's awkward for everyone

 

[Affiche]

Just as long as you don't start a thread dedicated to them complete with a poem and expressing your second life goal of marrying them and spending the rest of your life with them...that's awkward for everyone

Hahaha so true. Nope, I am persistently logical, not creepy/stalkerish. I did enjoy that thread though!

 

[Grace184]

Hahaha so true. Nope, I am persistently logical, not creepy/stalkerish. I did enjoy that thread though!

Haha I know you wouldn't do that! We always seem to comment on the same posts and be on par. But yea that thread was out of control!

 

[efle]

@gyngyn it's a slow season at work for me 🙂

I have a similar but slightly different chart. The difference is that I don't rank new MCAT scores to be above or below old MCAT scores, but equal to. Example: 30= 508, 509, not 508< 30< 509

The reason is that with the wider ranges of the old MCAT, you can't rank scaled scores between tests with varied percentiles. With the wide range of a 30, most 508s are better than a 30, but some 30s are still better than a 508. You're just looking at the ceiling of that range, and that's inaccurate. If a score, like a 508, falls within the range of another score, such as 30, they are equal. It's the same process of converting for the old MCAT: 75th percentile fell in the range of a 30, so it's equal to all other 30s.

Edit: should read ''a 508 is equal to or better than most 30s, but some 30s are still better than a 508''

The ranking system in this thread fails to factor in that some 30s were near the 73rd/74th percentile, which is lower than a 508.

You're misunderstanding the percentiles again. There are not high 30s vs low 30s. A 508 cannot be better than only some 30s, because all 30s are equivalent.

A 30 is higher than 73% of scores, and higher than or equal to 79% of scores (6% of testers scored equivalent 30s). A 508 is higher than 74% of scores, and higher than or equal to 77% of scores (3% of testers scored equivalent 508s).

So you can in fact rank new scores within old ones and vice versa, in this case using the ceilings (includes equal to percentiles).