More Statistics on the MCAT – It’s Harder Than You Think

I’ve scoured the internet trying to find an authoritative source for relating MCAT scores to percentiles and have never been able to.  However, this afternoon I finally found some information on the shape of exam score distributions.  I then proceeded to blow way too much of my afternoon computing cumulative probability distributions.  My last post on MCAT statistics has been one of my most popular pages, so I figured I’d try it again.  Before I get into this one, I need to get two things out:

• I absolutely hate the latest version of MS Excel.  This shouldn’t have taken more than about 20 minutes and it took me two hours instead because of how slow it runs on my Mac (and I have 8 GB of RAM!) and the fact that all of the normal functions have been made completely non-intuitive.
• This little exercise confirmed for me that physicians should take calculus.

I’m going to try and see if I can find a way to upload an Excel sheet that has all of the calculations that I did.  I’ve included a table below that shows the relevant data and I’ve also shown the cumulative probability distributions for both the individual sections as well as the overall exam.

A few provisos:

• I’ve assumed a normal distribution which may or may not be accurate.
• Not every MCAT has the same statistics and these figures are based upon a sample size of one, so some variance is to be expected.
• I’m not a statistician, so the perspectives and conclusions I drew from this might not be warranted by the data.

Here are cumulative probability functions for each of the 3 sections on the MCAT.  I ignored the writing sample.

Here’s the same quantity for the cumulative score.

A couple of observations to be made here:

• Notice that about 20% of those taking the exam scored below an 18.  That means that 1 in 5 people that write the MCAT will wind up with a trio of 6’s or worse.  Ouch.
• The magic score to get a person considered by an adcom is usually considered to be a 30.  This ends up being around an 80% score, which is just outside the first standard deviation for the exam.
• Getting anything above a 35 is really hard.  To put this into perspective, only 50 people out of every 1,000 that take the exam will do this.
• A final thing I noticed was the rarity of +37 scores.  There aren’t many at all, so this serves as really good evidence that you can’t believe anyone, particularly the student doctor forums.

I also noticed that the difficulty in moving a score from a 30 to a 35 seemed easier (a 15% difference) than moving from a 25 to a 30 (a 30% difference).  I’m not really sure what to attribute that to – it might have to do with the fact that individuals at that range may only differ by a couple of incorrect answers or that there are a lot fewer people left in that range.

Anyway, the end result that I got from this was that doing well on the MCAT is a lot harder than people seem to give it credit for.  My goal is a 35, but looking at these numbers, it’s easy to be discouraged.