Mcat 2015

[km2kenne]

Hey everyone, 

 

I've been a MCAT prep instructor for a couple years now, and I recently have been running sessions on the changes coming to the MCAT starting in 2015. 

 

I'm in no way directly involved in the creation of the new MCAT, or any MCAT prep materials, but I thought I would offer myself up to answer any questions you might have. I'm not saying I know everything, but I'd love to help out if I can! 

 

Something I want to clarify right off the bat is that I know this change is scary for a lot of people. The MCAT is getting a lot longer, the topics on it are changing, and there is very little prep material available currently. What you need to know is that the things that make the MCAT the MCAT are not changing. It still tests the same skills: namely critical thinking, problem solving, etc. It's still passage based. It's still entirely multiple choice. The actual content of the MCAT has always been the easiest part, so the fact that more content is being added isn't as big of a change as it seems. The biggest change is actually going to be that everything tested will be directly applicable to the study of medicine. There won't be the typical MCAT questions of friction on an inclined plane, but instead there might be a lever arm question involving the insertion point of the biceps relative to the elbow. 

 

There are still a lot of unknowns about the minutia of the new MCAT, but don't worry! There's still a lot of time and everyone else is in the same shoes as you. 

 

For content, I recommend checking out the MCAT 2015 content guide: https://www.aamc.org/students/download/377882/data/mcat2015-content.pdf

[BizMarkie]

Hey everyone,

 

I've been a MCAT prep instructor for a couple years now, and I recently have been running sessions on the changes coming to the MCAT starting in 2015.

 

I'm in no way directly involved in the creation of the new MCAT, or any MCAT prep materials, but I thought I would offer myself up to answer any questions you might have. I'm not saying I know everything, but I'd love to help out if I can!

 

Something I want to clarify right off the bat is that I know this change is scary for a lot of people. The MCAT is getting a lot longer, the topics on it are changing, and there is very little prep material available currently. What you need to know is that the things that make the MCAT the MCAT are not changing. It still tests the same skills: namely critical thinking, problem solving, etc. It's still passage based. It's still entirely multiple choice. The actual content of the MCAT has always been the easiest part, so the fact that more content is being added isn't as big of a change as it seems. The biggest change is actually going to be that everything tested will be directly applicable to the study of medicine. There won't be the typical MCAT questions of friction on an inclined plane, but instead there might be a lever arm question involving the insertion point of the biceps relative to the elbow.

 

There are still a lot of unknowns about the minutia of the new MCAT, but don't worry! There's still a lot of time and everyone else is in the same shoes as you.

 

For content, I recommend checking out the MCAT 2015 content guide: https://www.aamc.org/students/download/377882/data/mcat2015-content.pdf

For those wanting to score as high as possible (well, I guess everybody), what should our score goal be? On the "old" MCAT, the score I had hoped to get was around 35 (though I never wrote it as I chose to wait for the 2015). What would be the equilivant of that score be on the new MCAT? What are most schools going to consider competitive?

[daftypatty]

I believe there's only going to be one official FL exam from aamc out for the people taking the 2015 mcat next summer. How do you think these people should go on about practicing for the mcat in this case?.. There are FLs from Princeton and Kaplan, but I've always heard the aamc ones were best representative of the actual ones. Do you think doing the non-2015 aamc FLs would help? Or would those be a waste of time since the format is changing. 

[km2kenne]

For those wanting to score as high as possible (well, I guess everybody), what should our score goal be? On the "old" MCAT, the score I had hoped to get was around 35 (though I never wrote it as I chose to wait for the 2015). What would be the equilivant of that score be on the new MCAT? What are most schools going to consider competitive?

 

This is actually still quite up in the air. Since no one has written the MCAT yet it's hard to say what the "competitive" score will be in terms of scaled score because there is no scale. And they're changing how scores are scaled. For those who don't know, the new MCAT is scored between 118-132 in each section. Currently it's pretty close to impossible to score at the extremes of the 1-15 scale. Almost no one scores a 1 and almost no one scores a 15. Also, it's pretty useless to differentiate between someone scoring a 1, 2, 3, 4, or 5 at the low end because any of those scores is way too low, and ditto for scores of 15 or 14 or even 13 for the most part at the high end of the scale because those are all high enough to be treated fairly equivalently. With the new MCAT they're weighting more towards the ends of the scale so many more people will score the lowest score, and many more will score the highest score. 

[km2kenne]

I believe there's only going to be one official FL exam from aamc out for the people taking the 2015 mcat next summer. How do you think these people should go on about practicing for the mcat in this case?.. There are FLs from Princeton and Kaplan, but I've always heard the aamc ones were best representative of the actual ones. Do you think doing the non-2015 aamc FLs would help? Or would those be a waste of time since the format is changing. 

The old AAMC MCATs are still going to be useful. I'd recommend using the old AAMCs for the purpose of MCAT question simulation and alternative new FLs to practice mental stamina. Alternatively you could combine old MCATs and do two in a row, but that's not going to be a good simulation of the new MCAT for the material being covered. 

[MathToMed]

This is actually still quite up in the air. Since no one has written the MCAT yet it's hard to say what the "competitive" score will be in terms of scaled score because there is no scale. And they're changing how scores are scaled. For those who don't know, the new MCAT is scored between 118-132 in each section. Currently it's pretty close to impossible to score at the extremes of the 1-15 scale. Almost no one scores a 1 and almost no one scores a 15. Also, it's pretty useless to differentiate between someone scoring a 1, 2, 3, 4, or 5 at the low end because any of those scores is way too low, and ditto for scores of 15 or 14 or even 13 for the most part at the high end of the scale because those are all high enough to be treated fairly equivalently. With the new MCAT they're weighting more towards the ends of the scale so many more people will score the lowest score, and many more will score the highest score.

 

But in that case, it would no longer be (roughly) a normal distribution, which would cause calculation issues when trying to normalize the scores.

 

Have you heard that on good authority? Not that I don't believe you, but I just thought normalization was an improvement over a linear scoring system and it feels like a step backward.

 

I had (likely incorrectly) assumed that the 118-132 is identically distributed (118 = 1, 119 = 2, ... 132 = 15). The reason for the drastically different numbers being to quickly identify which test was taken (ie a 12 is clearly from the earlier test, a 130 is clearly from the new one etc. That would work for the amalgamated scores as well.

[SkepticalStudent]

For those wanting to score as high as possible (well, I guess everybody), what should our score goal be? On the "old" MCAT, the score I had hoped to get was around 35 (though I never wrote it as I chose to wait for the 2015). What would be the equilivant of that score be on the new MCAT? What are most schools going to consider competitive?

If you look at the AAMC's page on scoring, it looks like a competitive score may be around 510ish...? (Yep... guessing.)

 

I TAKE THAT BACK. 

 

From one of the AAMC guides:

 

n setting the score scales for the new exam, one goal is to re-center the distributions and spread out scores across the entire range of possible scores. Average scores for most of the current test sections have drifted up over time. The new score scales will re-center the distributions and correct the “bunching” at the upper ends of the scales. This means that the distributions of test takers’ scores will look more like normal distributions.

 

ETA: More from their guide.

 

Because the score scales for the new MCAT exam are different from the current scales, you [adcoms] are being asked 

to “start over” in attaching meaning to the scores and developing conventions for selecting students. 

Faculty will not yet know what scores mean at their schools. They will not know what to expect from 

students who enter at different score levels and which scores will and will not be associated with success.

For applicants with scores from the new exam, you will have percentile ranks that show how your 

applicants compare to others who tested on the new exam. These percentile ranks will help you compare 

applicants who took the new exam. The percentiles ranks will help give meaning to the new MCAT 

scores. They will show how well applicants “stack up” in relation to test takers who also prepared for 

and took the new exam.

It looks like there is a new focus on using percentiles rather than scores. So you could score 500 overall but not actually be in the 50th percentile but the 40th.

[km2kenne]

But in that case, it would no longer be (roughly) a normal distribution, which would cause calculation issues when trying to normalize the scores.

 

Have you heard that on good authority? Not that I don't believe you, but I just thought normalization was an improvement over a linear scoring system and it feels like a step backward.

 

I had (likely incorrectly) assumed that the 118-132 is identically distributed (118 = 1, 119 = 2, ... 132 = 15). The reason for the drastically different numbers being to quickly identify which test was taken (ie a 12 is clearly from the earlier test, a 130 is clearly from the new one etc. That would work for the amalgamated scores as well.

 

From my understanding, the scaled scores are still going to be normalized, but the tails of the normal curve are going to be truncated. That's from the AAMC guide quoted above saying they're going to be spreading scores across the entire possible range. 

[MathToMed]

Thanks for the info km2kenne. The remainder of this post isn't targeted at you specifically, I'm simply going to be mulling over the whole process.

 

------------

The issue I'm having as a math guy is that this would necessarily "denormalize" the curve even further. A normal distribution should have about 65% of the data falling between one standard deviation above/below the mean, about 95 between 2, etc. working its way outward toward the tail end.

Now if you look at the current MCAT's data tables, they do not match this normalized data - but they come fairly close if you assume a mean score of 8, and a standard deviation of 2.5 points. Technically, it is a "right skewed normal distribution." They allow for some variation, typically score inflation, in the event of a difficult section, and this is simply a flaw of having discrete data (ie. 32 / 40 marks) as opposed to interval/ratio data (like measuring a length/temperature).

 

See for example: https://www.aamc.org/students/download/361080/data/combined13.pdf.pdf

 

The way (I would assume) they calculate your score currently, would be to rank the test writers' scores in order, add a "difficulty score" based on the particular exam, and then calculate the median value, find the standard deviation/z-score of each test and assign a score appropriately. My understanding was that this was why they have normalized scores in the first place, as opposed to simply giving you a mark of 38/40, etc. Namely, it reduces a few sources of bias, by making the grading scale more coarse.

 

Think of a 4.0 GPA scale versus a percentile scale - the percentile scale is more accurate, but may vary considerably from school to school, course to course, because of a good/bad professor etc. whereas the GPA, while it may still vary "unfairly", does so to a far lesser extent. (This is an incomplete assessment of GPA and I realize that - a GPA score rewards consistency and penalizes spread more so than a percentile score).

 

------------

So the take home point - if they compress the curve, namely "squish down the middle" and "pull up the tail ends" (which is the same effect as truncating the ends) then first and foremost, they cannot use the same tried and true mathematical method of calculating student percentile rank. While I'm sure there are other valid ways of doing that (I'm not a stats major, though I've taken those associated with a mathematics degree), it leaves me wondering why they would make such a drastic change. If they wished to differentiate more, they should simply make the scale larger. That is, instead of giving a score of 1-15, give a score from 1-150. This dilation by a factor of 10 would be extremely transparent - even now I find myself wondering "was I at the high end of my VR score? Or the low end?" and I'm sure many others, including adcoms have had similar thoughts. Dilating by 10 would serve that precise purpose.

tl;dr - I have no clue why they're doing this haha.

[km2kenne]

I think they're not doing it to differentiate more, but rather to differentiate more only where it matters, and to make every score actually mean something. Right now I'd say there is no difference in meaning (whether you're competitive at any individual score) for scores from 3-20 overall (or 1-7 within any given section). Not sure on the mathematics of it though! 

[leemar3]

ANY HOME STUDY SCHEDULE FOR THE NEW MCAT?