MCATs, Affirmative Action, and DEI, Oh My

[Toni]

Ever heard of Occam's Razor?

I wonder what the acceptance rates were at the colleges that rejected his application? I wonder at the demographic composition of the class of incoming freshmen. My assumption is a significant number of Asian students were admitted at each of those schools. Also, first many/most state schools ( vs private) they accept only X% of out of state applicants.

We don’t know and never will know.

 

[thebeave]

Ever heard of Occam's Razor?

I wonder what the acceptance rates were at the colleges that rejected his application? I wonder at the demographic composition of the class of incoming freshmen. My assumption is a significant number of Asian students were admitted at each of those schools. Also, first many/most state schools ( vs private) they accept only X% of out of state applicants.

We don’t know and never will know.

For more head scratchers, you might also check out the extremely small subset of college applicants who have been accepted to all eight Ivy League schools.

 

[Toni]

Ever heard of Occam's Razor?

I wonder what the acceptance rates were at the colleges that rejected his application? I wonder at the demographic composition of the class of incoming freshmen. My assumption is a significant number of Asian students were admitted at each of those schools. Also, first many/most state schools ( vs private) they accept only X% of out of state applicants.

We don’t know and never will know.

For more head scratchers, you might also check out the extremely small subset of college applicants who have been accepted to all eight Ivy League schools.

I know it is an extremely small set—and should be. I’m surprised it even exists. Ivies are notoriously proud of their unique special schools and I would think they’d reject as being unserious anyone they knew applied to all the Ivies…

 

[Loren Pechtel]

So, how can we confidently discuss these as normal distributions when we can mathematically see there are skews?

Yeah, you don't have a normal distribution if you're not at the peak of the curve.

This gets even more complicated when we introduce confounding variables. For example:

• Socioeconomics: Research shows that factoring in parental education and family income virtually eliminates racial differences in MCAT scores.
• Preparation Capacity: Total hours dedicated to test prep is a strong predictor of MCAT success, which correlates with the financial capacity to afford extra educational resources and the free time to use them.

This raises bigger questions. When we group distributions purely as a function of race, are we actually looking at multi-modal, distinctly non-normal distributions driven by class and income?

Ultimately, this only underscores your broader point: we really do not know as much as we think we do from these tables, and people are making assumptions based on flawed, uni-dimensional models.

This part shows what's going on--as always, race is just a proxy for background. The problem is there's no reason to think that background doesn't have the same effect on medical school as it has on medical school admissions.

 

[Loren Pechtel]

It looks like I am mistaken about this section because the percentiles released are reflective of total test takers, not only matriculants. I am still looking into this and trying to confirm and understand what kind of distribution would best fit the data.

We are looking at one slope of a bell curve with clipping. I'm not aware of any model that can be standardly applied.

 

[Loren Pechtel]

This site is interesting:

Does MCAT Score Predict Success? Research Reveals Answer

Does MCAT score predict success? Explore research on exam predictability and discover how our MSMS program offers an alternative path to medical school success.

msms.tiberhealth.com

MCAT Success Doesn’t Seem to Translate to Clinical or Residency Success​

A study published in Military Medicine in 2015 looked at whether students with high MCAT scores were more likely to be evaluated as having strong clinical performance and professionalism later in medical school. The researchers looked at the GPAs, USMLE scores, and overall evaluations of students at the F. Edward Hébert School of Medicine, Uniformed Services University (USU) of the Health Sciences in 2010 and 2011. They found that high MCAT scores were:

• Associated with success on the USMLE step 1, the USMLE step 2 clinical knowledge exam, and the USMLE step 3
• Somewhat linked to higher GPAs during medical school
• Not consistently associated with success in clinical-based evaluations during the last two “clerkship” years of medical school, including the USMLE step 2 clinical skills evaluation
• Not associated with success on clinical evaluations during the first year of residency after graduating from medical school

The big title in bold is not consistent with the last underlined bullet quote, as the bullet quote says,"...during the first year of residency...". There are still several years of residency after the first year, and those are arguably more important for success as a full fledged doctor. Also, the bolded title makes a claim that "MCAT success doesn't seem to translate to... residency success", but that is not supported by any of the bullet points.

The study is from just one medical school and and during a two year period, so not really that comprehensive. It is odd though, why the first year is an exception.

Good catch. Yeah, this is an example of lying with statistics. They cherry picked the stuff that didn't show significance.

 

[Toni]

It is very possible to be an excellent test taker —and a less talented professional in any field you care to name.

 

[Loren Pechtel]

You are framing PGY-1 as an "exception," but logically, why would we expect MCAT scores to suddenly start predicting clinical success in Years 2–5 of residency if they fail to do so in Year 1?

P-hacking works both ways.

 

[southernhybrid]

Medical schools don't have room for, nor should they be expected to lower standards for any person.

In generalities, I agree. But I think there may be room for some squish that you're not taking into consideration. For example, it could be that in the past the bar has been set very high, as a means to limit the number of doctors, and thereby command higher rates. That's great... until we end up in a situation where there aren't enough doctors for the population and access becomes a problem.

If the bar is set higher than is required, then lowering that bar may be appropriate if doing so doesn't jeopardize quality.

Potentially timely analogy: It's all well and good to have your toilet paper bar set at soft, thick, doesn't pill, doesn't tear, doesn't have a nap when supplies are abundant. But when supplies run low, you might very well decide that all you actually care about is whether or not it will block your pipes.

Some people do better on standardized tests regardless of their intelligence and educational background and if anyone did a little research, they would realize as has been pointed out a few times, that MCAT scores do not make a good or a bad doctor. Things like emotional intelligence, willingness to keep learning, being open minded as new medical discoveries take place are far more important than MCAT scores.

Plus, a a nurse who has dealt with both good and horrible doctors, imo, many doctors don't bother to learn much of anything new once they begin practicing. This, imo is very true of while male doctors, not all of them of course, but far too damn many of them. Gotta go. Have fun with this silly argument or find something better to do with your time.

To add to this: for many undergrad students who are aiming for medical school, there is a strong competition for top scores, top grades, the most rigorous ( you name it) available—a kind of pseudo intellectual machismo.

This tends to produce students who are very very good at earning perfect scores but who do not necessarily develop good social skills, empathy, listening skills.

Social skills and empathy are nice to haves. Body of knowledge, ability to accurately diagnose and treat are foundational requirements.

So are social skills and some degree of empathy, depending on the field of medicine.

Patients do not follow the advice of doctors they do not trust. If they do not believe the doctor hears what they are saying and understands what they are saying, there is a break down in communication and the patient is less inclined to follow the doctor’s advice.

Not only that, but I will not continue going to a doctor who lacks empathy and knows how to relate to patients. I originally went to one doctor to have my knee replacement surgery, but didn't like his attitude. The one I ended up with was kind, and considerate and had a much better reputation than the first one. I do have some mild to moderate pain in my knees which is very common, but a friend of mine who went to the other doctor had much worse pain that made it impossible for her to exercise. Besides, I prefer NPs but they they don't do surgery. Sure, I've known one who was a jerk and didn't seem to be very bright, but she was the exception. I'm not a fan of PAs, as they often push unneeded things on patients. I've only known one who I liked.

I have not had much good luck with NPs except for one at the practice where I am now a patient, but I have had nothing but wonderful experiences with the PA’s I’ve seen or more specifically who saw my husband. Wonderful health care workers.

I had an NP tell me my pneumonia was allergies. I told her I'd been in the hospital two weeks prior for it and that the course of antibiotics was probably not long enough to get rid of it. I'd been taking thousands of milligrams of ibuprofen and chugging bottles of cough syrup to keep the cough down and the fever away, which I told her, but "No, it's just allergies." The fucking idiot told me to take Claritin.

About three weeks later, with a 103f fever, I staggered into the ER. I had a 1200 millimeter pleural effusion. The doctor asked me how in the hell I'd been walking around and working, and that's when I found out that if it burst through the lung wall that I would've drowned.

I spent ten days in the hospital and barely remember being there. I spent another month at home, but still wasn't right for a good six months.

Yay for NPs.

Sure, some are awful, but I've dealt with doctors who were just as stupid, including the one that got pissed off at me because I told him the drug he had ordered interfered with the patient's blood thinner. That could have killed her, but after getting pissed off, he stopped the drug the next day. Gee, maybe he should have looked that up before prescribing it.
Then there was the one I hung up on after he got angry because he wasn't making enough money being the doctor for low income pregnant women. I was reporting extremely high BP, a sign of preeclampsia, so I had to tell the patient to go to the ER.

That's the time I got a standing ovation from the hospital nurses when they found out I was the nurse who hung up on that doctor. And, btw, nobody made him delivery those poor women's babies. He had agreed to do it. Plus we had a nice semi retired doctor who did the exams prior to delivery.

I almost forgot about he ones who treated my rattle snake bite with a terrible experimental treatment that left my great toe fucked up to this day. Besides that the hospitalist never examined my foot while I was hospitalized and discharged me with a dirty dressing hanging from my foot, not to mention he never ordered an antibiotic or a tetanus vaccine. Luckily, I found an excellent doctor when I got back home and he was disgusted by how I was treated by the other doctors.

I could go on, but the point is that there are plenty of MDs who are idiots and plenty of NPs who are very smart and competent. I did see an NP once who wasn't very good, but I would never see her again. My former NP was much smarter than the doctor she had worked with and my current one is excellent, very compassionate and doesn't get upset when I refuse something. If you suspect a medical provider isn't competent, find a new one. No medical provider, regardless of their credentials is necessarily competent. It's usually not that difficult to figure out that they don't know what they're doing.

 

[southernhybrid]

It is very possible to be an excellent test taker —and a less talented professional in any field you care to name.

The opposite is also true. My son had terrible SAT scores, so he got a 2 year degree in computer technology with good grades and then went on to graduate with honors after completing his BS in computer science. He even built a computer from scratch for his kids. He's always had an outstanding talent regarding technology, despite not doing well on standardized tests or in high school. He's a programmer/developer with lots of experience now and was told he was the most valuable member of his team, when the asshole in charge started laying off federal workers. So, imo, standardized test scores don't tell you much about a person's abilities or talents one way or the other.

 

[Don2 (Don1 Revised)]

So, how can we confidently discuss these as normal distributions when we can mathematically see there are skews?

Yeah, you don't have a normal distribution if you're not at the peak of the curve.

This gets even more complicated when we introduce confounding variables. For example:

• Socioeconomics: Research shows that factoring in parental education and family income virtually eliminates racial differences in MCAT scores.
• Preparation Capacity: Total hours dedicated to test prep is a strong predictor of MCAT success, which correlates with the financial capacity to afford extra educational resources and the free time to use them.

This raises bigger questions. When we group distributions purely as a function of race, are we actually looking at multi-modal, distinctly non-normal distributions driven by class and income?

Ultimately, this only underscores your broader point: we really do not know as much as we think we do from these tables, and people are making assumptions based on flawed, uni-dimensional models.

This part shows what's going on--as always, race is just a proxy for background. The problem is there's no reason to think that background doesn't have the same effect on medical school as it has on medical school admissions.

I agree with the first part of what you are saying, but I don't think the "background" is entirely correlated to race. Preparation time is likely a little bit more independent than other variables, in my opinion. It's difficult to say for sure since we do not see statistics for this.

 

[Toni]

So, how can we confidently discuss these as normal distributions when we can mathematically see there are skews?

Yeah, you don't have a normal distribution if you're not at the peak of the curve.

This gets even more complicated when we introduce confounding variables. For example:

• Socioeconomics: Research shows that factoring in parental education and family income virtually eliminates racial differences in MCAT scores.
• Preparation Capacity: Total hours dedicated to test prep is a strong predictor of MCAT success, which correlates with the financial capacity to afford extra educational resources and the free time to use them.

This raises bigger questions. When we group distributions purely as a function of race, are we actually looking at multi-modal, distinctly non-normal distributions driven by class and income?

Ultimately, this only underscores your broader point: we really do not know as much as we think we do from these tables, and people are making assumptions based on flawed, uni-dimensional models.

This part shows what's going on--as always, race is just a proxy for background. The problem is there's no reason to think that background doesn't have the same effect on medical school as it has on medical school admissions.

I agree with the first part of what you are saying, but I don't think the "background" is entirely correlated to race. Preparation time is likely a little bit more independent than other variables, in my opinion. It's difficult to say for sure since we do not see statistics for this.

‘Background’ is a close proxy for socioeconomic class which is also correlated with education attainment of the parents.

This is why a lot of schools are placing a lower emphasis on SATs. Plus fewer college age prospective students.

Students whose parents graduated from university have a leg up on their fellow students, aside from the expected bump in family income: their parents know the system and are in a better position to help their offspring choose the correct classes in high school and college and are more likely to see value in and pay for prep classes.

Which also helps explain why elite schools admit students with slightly less impressive test scores/gpas if they are the first or first generation of students to attend college.

 

[Loren Pechtel]

We are not better off when people —often very good people—try to parse racist and bigoted ideas to tease out what might be some valid points, as if racism and bigotry held some kernels of truth.

Flip side of this is people trying to find racism in valid points.

They don’t.

But some people are so invested in the notion that white and ( certain) Asian people —especially males—are discriminated against because they are no longer first in line for all good things that they feel compelled to bring up MCAT scores in a totally unrelated discussion.

That’s pretty fucked up.

But some people are so invested in the notion that black and brown people--especially females--are discriminated against that they are unwilling to see the evidence. MCAT scores come up because it's a very apples-to-apples comparison with solid data.

 

[bilby]

Here's a snapshot of a part of a runprint from my work:



It's the header at the top of a detailed list of routes to be driven. Three pieces of information are provided: The sign-on time (quarter to three in the afternoon); The type of bus (a 14.5m long TAG-axle bus, most likely they will assign a Volvo B12); And the total distance to be driven (83.796km).

The latter information is provided to allow me to assess whether there is sufficient fuel to complete the trip. As part of my startup checks, I look at a variety of things that might cause me to reject the assigned vehicle - common examples being broken lights, flat tyres, or insufficient fuel to complete the assigned run.

It is my favourite example of spurious accuracy; The bus is 14.5m long, and may be collected from, and returned to, any location in a yard that is about 8,000m². Does the software that made that calculation really generate data accurate to the nearest metre? Do I care that the assigned work is 83.796km? Would 83.797km be too far? Do I look at the fuel gauge and think "I likely have sufficient diesel to get 83.795km, but 83.796km ain't going to be possible"?

Of course not. I guesstimate that a full tank will get me around 1,000km, and think "If there's less than an eighth, I will give it to the refuellers". That gives me more than 30km in reserve, so it should be OK, but I will keep a close eye on the fuel during the trip if the starting fuel is that low.

Basically only the first of the five digits provided is necessary; To me, 83.796km is "less than 90km", and a 30km reserve will cover most contingencies, including the imprecision of the fuel gauge, variations in road position en-route, and the unknown start and end point in the yard itself.

Am I being racist by disregarding the last four digits, and considering other factors, when I have been provided with data accurate to 6.89655% of the length of the vehicle? Is the data really that accurate?

If you analysed my fuel level go/no-go decisions, likely you will find some apparent bias towards higher or lower reserve fuel between Volvo vs Scania vs MAN buses. Is this evidence of discrimination?

There may well be some discrimination. But analysing these spuriously accurate figures cannot detect it for any given case, even if it is blatant.

113.54295 posts (approximately) so far in this thread, and all of them are a complete waste of time because they are debating whether or not an undetectable phenomenon exists.

 

[Loren Pechtel]

View attachment 55102

There are a couple ways to look at this.

1) OMFG!!! Blacks drop out about 250% more than whites!!! OMFFFFFGGFG!
2) Out of 1000 people, 977 whites don't drop out and 943 blacks don't drop out.

With these numbers, it makes one wonder, the importance of the MCAT other than being a basic screening tool.

The 250% is the relevant number. The other is being deceptive.

 

[bilby]

View attachment 55102

There are a couple ways to look at this.

1) OMFG!!! Blacks drop out about 250% more than whites!!! OMFFFFFGGFG!
2) Out of 1000 people, 977 whites don't drop out and 943 blacks don't drop out.

With these numbers, it makes one wonder, the importance of the MCAT other than being a basic screening tool.

The 250% is the relevant number.

Why?

The other is being deceptive.

How?

 

[Loren Pechtel]

I’ve talked with enough admissions counselors to know that admissions seeks to built a class that does not reflect the same very narrow world that most incoming first year students are accustomed to—but to bring together people of diverse backgrounds and experiences and talents. Because the world is a big wonderful place full of of many different kinds of people and that their talents, skills, ambitions, and abilities are not allocated by race, gender, sexual orientation ration/expression , class, ethnicity, first language or country where they or their parents or grandparents were born.

In other words, politically correct for discriminate in favor of minorities.

 

[Loren Pechtel]

You are all digging around in the noise, looking for a signal that can't possibly be there.

Test scores presented to three significant digits are a great example of spurious accuracy.

A sensible system would just give everyone with an MCAT above 505 a "pass" grade, and not bother to reveal the exact score, on the basis that exact scoring is not an exact representation of anything useful.

The idea that these results measure some dependable quality to the level of detail suggested by the use of three digit scores, rather than just saying (for example) that everyone who scored above 505 gets an "A", and anyone over 515 gets an "A+", and recognising that the difference between a score of 514 and a score of 516 is utterly insignificant, and that even bracketing students into two pools is therefore arbitrary (and bracketing them into dozens of scores, each one point apart, is both arbitrary and futile).

1) This is a standard dodge to hide discrimination.

2) You can't have 3 significant digits on an item with only 57 possible values.

 

[Loren Pechtel]

Even when it doesn't make sense--look at the flap about redlining around the turn of the century. Really, now, you expect me to believe bankers--all bankers--are only discriminating against blacks in black neighborhoods who want low down mortgages???

So to be a "pure scientist" you have to just straight up ignore all evidence that's in front of you? Throw out any data that match your bullshit rhetoric? That's science?

Redlining was LAW. Black-and-white, clear as day. It's only "disputed" now by people who are just as passively racist now as they were when they were letting it happen in the first place.

I'm talking about the findings around 2000, not the stuff from long ago.

 

[Loren Pechtel]

The local situation is much more sensibly explained by hypothesizing

You are unspeakably confused about the very nature of the scientific method. Like, at a primary school level. It doesn't matter what followed that phrase, nothing could have made that sentence scientifically sound.

Nothing wrong with my understanding of science. What's wrong is with your understanding of what's been proven.

The observed data was banks were less likely to write low-down mortgages in black neighborhoods. That doesn't prove why. They automatically assumed this was discrimination without attempting to investigate. But let's look at the possibilities.

1) Discrimination. But explain why all the banks do it (none sees an underserved market to exploit), explain why they only do it on low-down mortgages (why is there no such evidence of discrimination against other blacks.) No effort was even made to explain why this pattern existed.

or

2) Banks don't like underwater loans. The map of "discrimination" was basically the same map of price appreciation. Add a factor to the equation: what are the odds this mortgage will be underwater a few years down the road? Simple, makes good business sense, explains everything.

Which is more likely to be correct?