new MCAT trending high?

This forum made possible through the generous support of SDN members, donors, and sponsors. Thank you.

p0gono

Full Member
7+ Year Member
Joined
Jul 24, 2015
Messages
589
Reaction score
488
I recently saw this post below from the JHU admissions facebook. I was surprised and thought back to a recent comment I read/heard somewhere (from Pitt?) that said new and old MCAT distributions were looking about the same this year.

Thoughts? Is this perception going to hurt us who took the new MCAT and are applying this cycle?


I am back from a wonderful Annual Meeting of the AAMC. A few thoughts/observations:

- I personally sensed that the new MCAT scores seemed to be trending high. A number of my colleagues at other schools stated they have seen the same thing. The MCAT folks said this is because the applicants who took the old MCAT and then the new one are going to be more familiar with the exam, even though it is different. Of course, they were basing their data on an N of 2.

- Getting completely comfortable with the new MCAT will take time.

- Many medical schools seem to be up in application numbers (one school mentioned having a 26% increase from last year). While it is too early to know what this means, some wonder if the higher MCAT scores are leading applicants to apply to more schools to hedge their bet, and hopefully increase their chances.

 
Well if that is the case then thank god I applied this cycle before everyone knew. But how can it be trending high? Isn't the Mcat scored on percentile so only a certain amount of people can get a certain score based on how much they did better than the rest of the other test taker? Idk much about the MCAT but that's the perception I had
 
The new MCAT's scaled scores at the upper ranges correspond to lower percentiles than the old MCAT. Which is why 525+ scores have been reported with some frequency on SDN whereas 40+ scores on the old MCAT were quite rare.
 
I don't know of many that took the old one and the new one. Most people raced to get it done in January. In our May 22 MCAT group we didn't have repeat test takers very often.
I will wait until the data comes out next year. My score was good, but I wonder if it wouldn't have been higher without the biochem emphasis. If I could have concentrated on fewer subjects I imagine my scores would have been higher. The sample tests and question packs I took to practice were easier than the actual MCAT.
 
The new MCAT's scaled scores at the upper ranges correspond to lower percentiles than the old MCAT. Which is why 525+ scores have been reported with some frequency on SDN whereas 40+ scores on the old MCAT were quite rare.
+1
I have heard this from some faculty on the committee board also. They said those types of things come with the territory of a new test and over a course of a few years the percentiles will probably be adjusted.
 
I recently saw this post below from the JHU admissions facebook. I was surprised and thought back to a recent comment I read/heard somewhere (from Pitt?) that said new and old MCAT distributions were looking about the same this year.

Thoughts? Is this perception going to hurt us who took the new MCAT and are applying this cycle?

Yes, the new MCAT is going to kill you and eat your babies.

The score distributions in each section of the new MCAT are currently more normal than the old MCAT, at least at time of its retirement. The intended, idealized mean for each section on the old test was about 8, but the actual mean had crept up to over 9. Essentially the distribution had skewed to the right, with 10 being the mode. I believe the more normal distribution, combined with a finer scoring scale, will make it a bit easier to parse applicants with the new test.

There was some detectable disappointment, however, that the score distributions by sex and race had not changed between old and new exams. The mean for men was higher than women (501 vs. 498), and whites and Asians outperformed all other racial/ethnic groups by about same margins as before.
 
Last edited:
There was some detectable disappointment, however, that the score distributions by sex and race had not changed between old and new exams. The mean for men was higher than women (501 vs. 498), and whites and Asians outperformed all other racial/ethnic groups by about same margins as before.

Are there really people in education who buy into the cultural bias theory?
 
Are there really people in education who buy into the cultural bias theory?

In medical education? Sure, but it's less of an issue than at the undergraduate level. Cultural bias has been an extremely controversial topic with regard to the SAT, for instance. In graduate/professional education it is much more common to ascribe apparent race-based performance variations to stereotype threat.
 
As a poster said above, I bet this is purely perceptual and for two reasons. Firstly regarding percentiles - JHU sees a lot of far high end applications, and the "max" reported for the higher end is now the top 0.5% (523+ reported as 100th percentile) instead of the top 0.1% (41+ reported as 99.9th percentile). So perceptually the number of perfect percentiles increased by a factor of 5. Secondly regarding raw scores - subsections had higher end scores collapsed into a single new score, for example 13, 14, and 15 in Verbal are all now a 132 CARS. If an adcom looks at scores in terms of # points shy of max, they'll be misled by this into thinking things like 526 = 43 or 131 = 14, would very easily lead to faulty perception that MCAT scores are trending much higher at the upper end.

Are there really people in education who buy into the cultural bias theory?
As opposed to what alternative?
 
Yes, the new MCAT is going to kill you and eat your babies.

The score distributions in each section of the new MCAT are currently more normal than the old MCAT, at least at time of its retirement. The intended, idealized mean for each section on the old test was about 8, but the actual mean had crept up to over 9. Essentially the distribution had skewed to the right, with 10 being the mode. I believe the more normal distribution, combined with a finer scoring scale, will make it a bit easier to parse applicants with the new test.

There was some detectable disappointment, however, that the score distributions by sex and race had not changed between old and new exams. The mean for men was higher than women (501 vs. 498), and whites and Asians outperformed all other racial/ethnic groups by about same margins as before.
As a poster said above, I bet this is purely perceptual and for two reasons. Firstly regarding percentiles - JHU sees a lot of far high end applications, and the "max" reported for the higher end is now the top 0.5% (523+ reported as 100th percentile) instead of the top 0.1% (41+ reported as 99.9th percentile). So perceptually the number of perfect percentiles increased by a factor of 5. Secondly regarding raw scores - subsections had higher end scores collapsed into a single new score, for example 13, 14, and 15 in Verbal are all now a 132 CARS. If an adcom looks at scores in terms of # points shy of max, they'll be misled by this into thinking things like 526 = 43 or 131 = 14, would very easily lead to faulty perception that MCAT scores are trending much higher at the upper end.


As opposed to what alternative?

But would you say the higher scores in the new MCAT seem more believable as opposed to straight up luck?
 
As a poster said above, I bet this is purely perceptual and for two reasons. Firstly regarding percentiles - JHU sees a lot of far high end applications, and the "max" reported for the higher end is now the top 0.5% (523+ reported as 100th percentile) instead of the top 0.1% (41+ reported as 99.9th percentile). So perceptually the number of perfect percentiles increased by a factor of 5. Secondly regarding raw scores - subsections had higher end scores collapsed into a single new score, for example 13, 14, and 15 in Verbal are all now a 132 CARS. If an adcom looks at scores in terms of # points shy of max, they'll be misled by this into thinking things like 526 = 43 or 131 = 14, would very easily lead to faulty perception that MCAT scores are trending much higher at the upper end.


As opposed to what alternative?


I think the alternative is that the problem isn't the testing but for a myriad of other reasons blacks score lower than whites and asians.
 
As opposed to what alternative?

If socioeconomic differences have already been accounted for and a racial disparity still exists, then cultural practices, values, and norms regarding education and achievement are still more likely to explain the discrepancy than biases present in the test itself.
 
But would you say the higher scores in the new MCAT seem more believable as opposed to straight up luck?
I'm not sure I follow. I think a 132 and 13+ are equally "believable". In neither case is there much luck involved, some people can hit that mark repeatedly across practice tests and explain their reasoning behind each answer.
 
I'm not sure I follow. I think a 132 and 13+ are equally "believable". In neither case is there much luck involved, some people can hit that mark repeatedly across practice tests and explain their reasoning behind each answer.

Since a 13-15 in VR collapsed into 132 in CARS, it'd probably suggest that the CARS section is more forgivable to error than the old VR section, resulting in a reduction of randomness/luckiness. So this'd make the new MCAT overall more reliable/less error-sensitive than the old MCAT.
 
cultural practices, values, and norms regarding education and achievement
Just to clarify, this argues an under repped minority premed on test day will likely underscore an Asian counterpart because they've been raised to just not care as much about the test?
 
How are you making that assumption?

Threads get derailed way to much on this site so I'll be short. Blacks, for many reasons, have been shown to underperform in virtually all areas of academic achievement. Ask yourself what is more likely, that all tests everywhere are racist against black and for whites/asians, or that there is some underlying variable already extant within the black population that causes this underachievement is all fields.
 
Just to clarify, this argues an under repped minority premed on test day will likely underscore an Asian counterpart because they've been raised to just not care as much about the test?

That's a gross oversimplification and you know it.
 
Wait see the edited post.
Agree again. They really should've just kept the Verbal reporting as 13-15. The fact that two verbal questions could equate to +/- 2 on your composite was ridiculously oversensitive
 
Agree again. They really should've just kept the Verbal reporting as 13-15. The fact that two verbal questions could equate to +/- 2 on your composite was ridiculously oversensitive

But isn't that the issue of fewer questions? What is surprisingly interesting is before the CBT shift of the MCAT, the paper exam was actually longer and less sensitive to errors! So the new CARS section having more questions (as well as the new MCAT as a whole being a lot longer), it's actually a good thing and more reliable as a metric in the end!

@gyngyn @Med Ed and others. Do you know why the paper exam was reduced to a shorter, but harsher exam when it switched over to CBT?
 
stereotype threat, home environment/ses are a couple
SES controlled ofc. Stereotype threat is an interesting one since afaik it can do a good job of explaining low performance from threatened minorities but can't explain overperformance by some other groups (eg Ashkenazi Jewish outperforming Caucasian). Strikes me as clearly a big part of the answer, but only a part.
 
But isn't that the issue of fewer questions? What is surprisingly interesting is before the CBT shift of the MCAT, the paper exam was actually longer and less sensitive to errors! So the new CARS section having more questions (as well as the new MCAT as a whole being a lot longer), it's actually a good thing and more reliable as a metric in the end!

@gyngyn @Med Ed and others. Do you know why the paper exam was reduced to a shorter, but harsher exam when it switched over to CBT?
Good for the evaluators, but must be a terrible experience for the testers!
 
SES controlled ofc. Stereotype threat is an interesting one since afaik it can do a good job of explaining low performance from threatened minorities but can't explain overperformance by some other groups (eg Ashkenazi Jewish outperforming Caucasian). Strikes me as clearly a big part of the answer, but only a part.

Ashkenazi Jews outperforming run-of-the-mill Caucasians can be attributed to stereotype boost.
 
Ashkenazi Jews outperforming run-of-the-mill Caucasians can be attributed to stereotype boost.
There would have to be a negative stereotype about white tester performance for that to work, which there isn't. White males are the classic example of people who don't ever have poor expectations foisted on them
 
There would have to be a negative stereotype about white tester performance for that to work, which there isn't. White males are the classic example of people who don't ever have poor expectations foisted on them

No, that's stereotype lift. Stereotype boost only requires positive stereotypes about one group.
 
No, that's stereotype lift. Stereotype boost only requires positive stereotypes about one group.
Ah, never noticed that distinction before, thanks. But again doesn't really exist; eg society doesn't project "good at IQ tests" onto Ashkenazi Jewish
 
Maybe it's that Ashkenazim have high IQs. Not saying that's necessarily genetic but aren't we doing gymnastics to justify something that already has a justification? That one average Ashkenazi Jews are simply smarter than other groups, followed by East Asians and Northern Europeans? We all can look at how/where the average white Jew is raised versus how/where the average black kid is raised. Aren't stereotype boosts/lifts/threats superfluous?
 
Ah, never noticed that distinction before, thanks. But again doesn't really exist; eg society doesn't project "good at IQ tests" onto Ashkenazi Jewish

I don't know about society as a whole, but I have been aware for many years of research indicating that Ashkenazi Jews are the highest-performing ethnic group at IQ tests. But IQ or standardized tests are usually written by WASPs, so cultural bias cannot explain it either. If you don't think stereotype boost is the answer, then it has to be either genetics or cultural attributes.
 
But IQ or standardized tests are usually written by WASPs

What limits that explanations is that WASPs don't actually do all that well on IQ tests compared to Jews, Northern Europeans, and East Asians.
 
Maybe it's that Ashkenazim have high IQs. Not saying that's necessarily genetic but aren't we doing gymnastics to justify something that already has a justification? That one average Ashkenazi Jews are simply smarter than other groups, followed by East Asians and Northern Europeans? We all can look at how/where the average white Jew is raised versus how/where the average black kid is raised. Aren't stereotype boosts/lifts/threats superfluous?

I think the testing differences in race/background come primarily from how the children are raised; what values and behaviors are instilled in them. It also helps to have two parents at home instead of just one. The average African American child lives in a single parent household (a whopping 67% in 2013). See: http://datacenter.kidscount.org/dat...6,868,867,133,38/10,168,9,12,1,13,185/432,431
 
I don't know about society as a whole, but I have been aware for many years of research indicating that Ashkenazi Jews are the highest-performing ethnic group at IQ tests. But IQ or standardized tests are usually written by WASPs, so cultural bias cannot explain it either. If you don't think stereotype boost is the answer, then it has
to be either genetics or cultural attributes.
Which brings us full circle to what's an example of how a cultural norm can make me score higher on an exam?

I think the testing differences in race/background come primarily from how the children are raised; what values and behaviors are instilled in them. It also helps to have two parents at home instead of just one. The average African American child lives in a single parent household (a whopping 67% in 2013). See: http://datacenter.kidscount.org/dat...6,868,867,133,38/10,168,9,12,1,13,185/432,431
The stuff we're referring to is controlled for SES singleparent etc
 
Guys why does this always happen
 
Which brings us full circle to what's an example of how a cultural norm can make me score higher on an exam?

Social group practices. If your friends are all taking MCAT prep courses or hiring tutors, then you are more likely to do the same. And of course, we know without a doubt that people tend to associate along racial lines.
 
As a poster said above, I bet this is purely perceptual and for two reasons. Firstly regarding percentiles - JHU sees a lot of far high end applications, and the "max" reported for the higher end is now the top 0.5% (523+ reported as 100th percentile) instead of the top 0.1% (41+ reported as 99.9th percentile). So perceptually the number of perfect percentiles increased by a factor of 5. Secondly regarding raw scores - subsections had higher end scores collapsed into a single new score, for example 13, 14, and 15 in Verbal are all now a 132 CARS. If an adcom looks at scores in terms of # points shy of max, they'll be misled by this into thinking things like 526 = 43 or 131 = 14, would very easily lead to faulty perception that MCAT scores are trending much higher at the upper end.

If they used something like your conversion table, this should be corrected for, right?
 
Social group practices. If your friends are all taking MCAT prep courses or hiring tutors, then you are more likely to do the same. And of course, we know without a doubt that people tend to associate along racial lines.


Which brings us full circle to what's an example of how a cultural norm can make me score higher on an exam?


The stuff we're referring to is controlled for SES singleparent etc

But they're not controlled for parenting.
 
It's really difficult to control for the socio element of socioeconomic status

Exactly -- and it plays a major role in how successful children will ultimately be.
 
This is clearly moving off topic, but....Just a few points, will try to cite later but have covered these in various cultural neuropsych courses:

-There are racial/cultural differences in expectations of milestones and how much parents push children to these milestones. There are American studies on spatial/coordination skills etc that Indian> whites > blacks > Puerto Ricans in their expectations for development.

-There are racial/cultural differences in how parents view infants and young children. For example as a little person, capable of understanding everything or as a baby that is fully dependent and virtually clueless. These differences change how much they speak and interact with the child as an infant and through the critical years.

-The differences in how much an infant is spoken to, leads to a 3 million word difference in vocabulary and understanding by the time children are 9 years old from the time they develop language. Clearly at the elementary school level this is a large part of performance (spelling, reading, writing) and there is tons of evidence on how early performance predicts later test scores, the confidence, study skills, continued support etc. The same type of advantage can be seen based on SES status, but studies have specifically looked at how parents interact with children.

Test scores remain fairly stable over time. And then there is all the categories of stereotype threat and lift among other cultural factors.
 
Which brings us full circle to what's an example of how a cultural norm can make me score higher on an exam?


The stuff we're referring to is controlled for SES singleparent etc
this is a guess but collectivist vs individualistic--> level of parental involvement
 
Top