MCATs, Affirmative Action, and DEI, Oh My

[Derec]

You are making a few significant logical leaps here, primarily by conflating correlation with causation

I am making a logical inference from the available data. You are the one making leaps by trying to deny any significant role of academic metrics in failure in medical school.

and equating "dropping out" with "failing." Let's break down the variables and dimensions you are leaving out of this equation.
Retention vs. Failure: First, a 94% retention rate for the lower cohort is still overwhelmingly high.

It may be high, but 6% dropping out is certainly not negligible.

More importantly, you are equating "withdrawing" with "academic failure" (e.g., failing out due to poor grades).

Given the high sunk costs of attending medical school, not only the substantial tuition already paid, but taking the MCAT, prereq classes, and doing all the shadowing, volunteering and other subjective stuff adcoms look at, I do not think many medical students quit for non-academic reasons. Do you have any data showing otherwise?
Note also that MCAT and GPA is strongly correlated with academic success in med school, including massing Step 1 and graduating on time as well as graduating within 5 years (i.e. up to one extra year). See here. So it is hardly surprising that if you admit a certain group of people with lower MCAT and GPAs in the name of "diversity" that they also have higher failure rates.

A massive confounding variable here is socioeconomic status.Wealth correlates heavily with higher MCAT scores,

Wealth always helps. But wealth correlates far more heavily with having parents who have connection in medical schools, or connections at research labs (to get their scion on some papers) or can hire a professional writer to write their personal statements and other essays medical schools want. MCAT is about knowledge and reasoning, and parental wealth matters less than for those subjective measures your Ilk wants med schools emphasizing even more than they do now.

and it also correlates with the ability to survive the severe financial strain of medical school. Assuming that 6% attrition is purely a lack of intellectual ability is a uni-dimensional assumption that ignores the reality of medical student burnout and financial attrition.

Maybe not purely, but largely. 24% (almost 1/4) of those with a 498-501 MCAT fail Step 1 on the first attempt. 8% don't ever pass and can't proceed in their medical education. For those with 502-505 MCAT it's 14% and 5% respectively. So I don't think it's a leap at all to say that academics is also the reason why the group with lower MCAT and GPA also has a 2.5x higher rate of dropping out. I think it's a much bigger leap to assume that it's mostly other factors, or that MCAT and GPA do not measure the level of academic preparedness for med school.

And again, what do you think is better than MCAT and GPA at measuring academic preparedness?

Tests Predicting Tests: You highlight that the MCAT predicts USMLE Step 2 CK. That is true to an extent, i.e. there is some kind of non-zero correlation, but it is also circular.

It also predicts passing Step 1, both on first try and ever. 8% of those with 498-501 and 5% of those with 502-505 MCAT never pass it. That means that they have wasted years of their lives and tens of thousands of dollars of tuition because of academics.

Standardized test-taking ability predicts future standardized test-taking ability. That isn't a revelation, and it certainly doesn't contradict the fact that actual medical practice requires more skill sets than the MCAT tests.

Of course it requires more skill. I think scrapping the clinical skills portion of Step 2 was a mistake.
But how would you assess aptitude for those skills on a college graduate applying for med school?

And medicine requiring more skill sets does not discount the importance of medical knowledge and reasoning, which is what USMLE assesses in standardized fashion. I.e. every MD student (and many DOs) take the same test, and are thus compared using the same yardstick.

The "Subjective" Dismissal: You wave away clinical and residency evaluations because they are "subjective." Subjectivity isn't a flaw here; it is the only viable mode of observation for hands-on, real-world medical practice.

I am not dismissing it, but pointing out that this is a major flaw of clinical evals and that we should not do away with even more objective measures in order to overly rely on these subjective ones. Every preceptor is different in how they evaluate their students, and evaluations can be affected by preceptor bias, personal connection or mood. And how about family connections? A student's father may be a physician who knows the preceptor, or knows somebody who knows the preceptor. Subjective measures are more susceptible to connections than an exam,

A multiple-choice exam cannot measure physical dexterity, social intelligence, bedside communication, or psychological resilience when dealing with patient trauma. The tests do not measure the multi-tasking nor real world on-the-job stress capacity either. Medicine is fundamentally a physical and social profession, in addition to knowledge.

But knowledge and reasoning matter a lot too. Other things you mention matter too (which is why again Step 2 CS should be brought back) but there needs to be a way to fairly assess these things. And that is the problem with subjective measures.

And why are we talking about rotations? This is about getting into med school. These premeds have not been on any rotations. They have not been given any incisions to suture. You can't base your admissions decisions on how they will have performed on their Year 3 rotations when they are still years away from it. Or do you think med school adcoms look like this?

You cannot simply hand-wave away some of the most critical dimensions to being a doctor just because they don't fit neatly into the uni-dimensional narrative you have constructed.

First of all, I am not arguing for a unidimesnsional narrative. I am just saying that all these things you listed are hard to assess in med school applicants. And writing essays and volunteering at a soup kitchen doesn't do much to assess somebody's dexterity either. So it's not even the case that medical school admissions are based on many of the metrics you listed. They may be assessed during rotations, but not for med school admissions.
Second, there is no reason to think that other metrics med schools use go the opposite way to MCAT and GPA. Because in absence of deliberate racial preferences, that would have to be the case.

The Statistical Trend: You are missing that the article shows a decaying relationship as training progresses and is backed by broad empirical data. In studies, the predictive validity (r value) of the MCAT to written boards like Step 1 is stronger. However, its correlation to clinical clerkships begins to plummet. The later lack of correlation to MCAT in 1st year of residency is part of the overall trend.

That the predictive value of MCAT/undergrad GPA wanes over time is hardly surprising. One reason is attrition. If you never pass Step 1, you never see rotations. Another reason is that of course, people continue developing over time. But why should that matter? Adcoms do not have access to their hypothetical future performance. They only have access to what the student has done so far, of which MCAT and GPA are a big part.
And if you say, well, we should give 502 MCAT scorers a chance anyway, then why does a 502 black student have a decent chance of US MD admissions, while an Asian with the same score has basically no chance? Do you think Asians on average lack social skills etc. compared with blacks, or is it more likely that medical schools practice racial preferences like they have done far more overtly in the days of UC Davis quota system?

So, once a student survives the classroom years, the variance in their success is driven by real-world physical and social dimensions. Trying to flatten the complexity of producing a good doctor down to a single test score is just not an intellectually rigorous way to look at the data. That does not mean MCATs are not valuable nor should be utilized early in the medical school process; it just means that once medical school is completed and the doctor has made it through all the testing and the whole process, the difference in scores is far less significant than other factors.

I never said other things do not matter.
But certainly two students with comparable applications should have a comparable chance at admissions. I see no reason why we should require Asians and whites to have better academic qualifications than blacks and Hispanics. And yes, the data clearly show that they do. Unless you want to claim that blacks on average have so much better essays and volunteering experiences that it compensates their lower MCAT/GPA averages. I find that very unlikely, and denying the obvious - that med schools practice racial preferences - is untenable. Especially since we know that academia, including med schools, have a long history or practicing and defending racial preferences.

 

[Derec]

Without any evidence that the resulting minority doctors are less successful in their practice of medicine than whites, your opinion appears more like bigoted sour grapes than evidenced-based disinterested and reasoned analysis.

The data we have (like dropout rates) suggest lower performance because so many marginal students are admitted because they have the politically correct melanin content and ancestry.

Do you have any data contradicting it?

 

[Derec]

Your statement is incredibly vague.

No, it isn't.

If you are comparing an Asian cohort to an African American cohort on a test with a hard 528 ceiling, the skew matters.

What is the skewness of these distributions anyway? Do you have any data on that? It is not reported by AAMC.

For a group with a higher mean, the right tail of their distribution crashes into that ceiling and compresses heavily (a left skew). For a group with a lower mean, their curve is much less restricted at the top. Treating these as identical "perfect normal distributions" mathematically distorts the upper percentiles, which are precisely the percentiles we are discussing when looking at medical school acceptances.

Again, I doubt very much the skewness matters for these distributions. But you are welcome to present data rather than speculation.

For the purposes of statistically comparing means, skewness does not matter at all. That is because for large Ns the distribution of the means becomes normal no matter the underlying distribution, with the Standard Error of the Mean being the standard deviation divided by square root of sample size. With thousands of people in most of these groups, SEM becomes very small.

Assuming normality across these aggressively filtered, socioeconomically stratified sub-populations is just that: an assumption.

Assuming significant deviation from normality is just as much an assumption.
But even if there was such a deviation, it does not matter when it comes to comparing means because of our good friend the Central Limit Theorem.

To somehow say that with thousands of people in each group there is no real difference between the group with a 505.7 vs. 514.3 is just solipsism. And if you really think that, you'd have to expect that in different years we'd see different groups having top MCAT and GPA averages. But that's not what we see. No, the relative distributions are pretty stable from one cycle to the next.

This is a false dichotomy. I never said I "hate the MCAT" or that we should abandon it.

Do you then think different racial and ethnic groups should have different MCAT and GPA standards?

The MCAT is a valuable tool early in the process for predicting who can survive the grueling didactic classroom years (Years 1 and 2). However, as I pointed out in my previous post, the empirical data shows that MCAT scores do not correlate with clinical success during residency. Once a student survives the classroom years, the variance in their success is driven by real-world dimensions: physical, real life multi-tasking, working with colleagues, resilience, dexterity, bedside communication....

But no med student gets to clinical years, let alone residency, without surviving the preclinical years and USMLE.
And how do you assess "physical, real life multi-tasking, working with colleagues, resilience, dexterity, bedside communication" in an undergrad applicant? Let alone, why would you think any of these metrics favors blacks and Hispanics over Asians and whites. Because if we assume "physical, real life multi-tasking, working with colleagues, resilience, dexterity, bedside communication" are evenly distributed across races, then the admitted students should also have similar MCAT/GPA distribution regardless of race, unless there are racial preferences.

Why are you and your Ilk so adamant in denying that those preferences exist and that they are the reason for the stark differences in MCAT and GPA distributions of accepted students by race/ethnicity? ZiprHead at least admits it, even as he is supporting it as long as the groups he dislikes get discriminated against.

Saying "the MCAT is not the sole, permanent predictor of a good doctor" is not the same as saying "the MCAT is useless."

Some people are going that far, just like with SATs. Nobody is saying that MCAT is the sole predictor of anything. But it is an important one, along with grades, of a student's academic preparedness. Accepting marginal students in order to increase "diversity" makes these students more susceptible to failing Step 1 and not graduating.

Again, you are looking at this uni-dimensionally. You are assuming the only valid metrics are standardized test scores and GPAs.

No, I am saying that MCAT/GPA are, as far as we know, the only metrics with marked differences between groups among admitted students.
If these other metrics are similar for e.g. blacks and whites, then the only thing that can explain the differences in objective academic metrics is discrimination.
Your insistence that it's not discrimination would be a little more understandable if not for long and well-documented history of med schools engaging in racial discrimination in order to shape their student body. Bakke vs. UC Regents was half a century ago, and med schools have continued using different methods in order to achieve what they want. Hell, University of Pittsburgh fired a fellowship director for being critical of "affirmative action".

Med school professor removed by UPMC as fellowship director over white paper

Different medical schools have entirely different missions and different requirements with different applying racial demographics. Many state and regional medical schools have explicit charters to train doctors who will practice in underserved rural or inner-city communities.

I know pretty much all students put that they want to pursue primary care in an underserved community in their essays for that reason. But would it not be better to have some sort of tuition forgiveness if you pursue primary care in the med school's community? Pretending that melanin and ancestors who spoke Spanish predicts interest in underserved primary care does not seem that well-thought out to me. Or the corollary that epicanthic folds and less melanin means predicts an interest in plastic surgery and dermatology in one of the top 5% zip codes in the country.

If an applicant has a slightly lower MCAT but a demonstrated history of working in those specific environments--which you dismissed as "volunteering at a soup kitchen"--that applicant is a objectively better fit for that specific school's mission.

But data do not show slightly lower MCAT and GPA but significantly lower scores. Among accepted students we have 514.3 vs. 505.7. That's not "slightly" anything.
And I am dismissing "volunteering at a soup kitchen" because it does not do a good job demonstrating somebody's genuine interest in helping others. Clinical and nonclinical volunteering has become expected and thus everybody does it. So it can't really be used to distinguish one applicant from another.

You are treating a chart of average MCAT differences as an absolute proof of systemic discrimination.

You certainly have not offered any reason as to why it doesn't. Especially given the well-documented history of med schools discriminating by race.

While there may be isolated instances of unfairness, the broader system is functioning as intended: committees weigh the MCAT heavily for academic baseline, but they also weigh other factors, such as the specific mission of the medical school.

MCAT and GPA predict higher chances of admission in every racial group. But those chances are also very different between groups.
This graph is older and uses the old MCAT format. I wish there was a newer version available, but alas, we make do with what we have.

As you can see, lower-performing blacks had about the same chance of admission as higher-performing Asians. Only 6% of low-performing Asians and 8% of low-performing whites got in, but more than half of low-performing blacks. Among high-performers, only just above half of Asians got in, but almost all blacks. I doubt that is explainable with blacks leaving Asians in the dust when it comes to "physical, real life multi-tasking, working with colleagues, resilience, dexterity, bedside communication".

And the data backs up this holistic approach: once these students actually make it to residency, their initial MCAT disparities no longer predict their clinical success.

That too does not explain why Asians and whites should be expected to have higher MCAT and GPA compared to Hispanics and blacks.

 

[Derec]

Those are anecdotal examples, not a consistent methodology, but In short, "honestly looking at race", to you, means carefully (and exclusively) ruminating on all possible offenses to the White race you imagine yourself to be a member of, and calling out any Black popular celebrity who offends Whites. Is that correct?

No.

Personally, to try and place myself on one side of a two-sided "race war" would mean antipathy toward a part of myself and my body, a larger part of my family, and most of my community. How could I live on such terms, always hating and being hated at all times? This war of yours... I think you will find that winning it would give you no lasting pleasure even if you succeeded.

Winning to me would mean defeating racists on both sides and looking at people as individuals. The Right may be beholden to group identity, but so is the Left, currently even more so than the Right.

 

[Politesse]

What "group identity" is the left "beholden to"?

 

[Derec]

Denial of the generational effects of racism.

Those cannot be fixed by perpetually discriminating against members of other groups.

Gospel is right that AA/DEI only addresses symptoms. But like any treatment of symptoms it is meant to be a temporary stopgap, since it does not fix the underlying cause. Just popping antipyretics without addressing the infection will not go well for you. Neither will taking corticosteroids without addressing the reason why they are used in the first place - just like AA, long term use has side effects.

 

[Derec]

What "group identity" is the left "beholden to"?

Blacks (vs. whites), women (vs. men), trans (vs. cis), Muslims (vs. Christians and Jews)

That's why no white person or man was allowed to be in the running for Biden's SCOTUS pick or running mate. That's why DNC decided to put SC on the pole position of the primary calendar even though the primary electorate is completely unrepresentative.

 

[Derec]

Why assume normality? More to the point, why assume a non-skewed distribution, when there's very good reason to think that skew likely exists?

Because of the large N. When we "assume normality" we don't mean perfectly normal, just close enough.

Realistically, I don't really care to continue this line of argument. I provided an illustration to demonstrate that a single threshold could be used that would produce very different mixes in outcome by raising and lowering for the entire population. And that was on the extremely simplistic assumption of skewlessness.

As I said, your example does not work. You assumed populations with very different standard deviations, when distributions of MCAT/GPA for different groups have similar standard deviations. You would not get the same result using real world data.

 

[Derec]

Uh huh...

About the level of discourse I have come to expect from you.

 

[Derec]

No it isn't. It's lying with statistics... even if it's unintentional. In reality, only 2.3% of whites drop out, and only 5.7% of blacks. The difference is not material.

2.5x chance of dropping out is certainly material, even if something is relatively rare (and 1 in 17-18 is not that rare either). If something increased your chances of getting cancer from 2.3% to 5.7% you'd be well advised not to ignore it.

This is the statistical equivalent of showing a vertical bar chart where the axis is set to 900 instead of 0 and saying

That's not the same at all.

"Look how big the difference is!" It's not big. It's not meaningful. It's noise.

Given the large N, it is definitely not noise.

 

[Derec]

Actually very few students drop out of medical school. Those who do almost always do fir personal or financial reasons, not academic ones

[citation needed]

 

[Derec]

It has more to do with the volume of doctors needed in the population. And reducing the bar for entry will produce more doctors.

Without increasing the number of residencies, it will not produce more practicing physicians. It will produce people like Abdulrahman El Sayyed - he finished med school, but never practiced.

If the bar was kept artificially high in order to command a higher price, but demand can no longer be met, then lowering the bar is a rational and appropriate response, provided the bar is still sufficient to guarantee quality.

Wherever the bar is set, it should be the same regardless of melanin or Spanish-speaking abuelos.

 

[Derec]

Actually, about 7% not 14%. The NBA is strictly males and black males are 6-7% of the population. You need to compare males to males.

As you said, NBA is just males. So the denominator is "all males" and numerator is "black males". So, 14% is correct.

 

[Derec]

Because they are still suffering from the racism their parents and grandparents experienced.

How long should this special treatment last in your opinion? I do not think there is any person alive today whose ancestors have not experienced something horrible. I do not think this is valid reason to treat individuals differently.

Two closely matched candidates for a position or scholarly program. Give it to the person whose parents or grandparents were discriminated against and forced the minority candidate to work harder for the position.

Under "affirmative action" candidates of certain minority groups can work far less hard for the position and still get it.
People should be treated as individuals. Past discrimination does not justify present discrimination.

 

[Derec]

Well, you sould say that, if you understood the statistics.

My example was purely hypothetical. You sould[sic] have understood that.

That was fewer than six incidents per million departures.

So a very different order of magnitude than slightly fewer than six dropouts per one hundred students. And that's after first year alone. Low MCAT/GPA gang also has a markedly higher Step 1 failure rate, and if you don't pass you can't continue to year 3.

Rest snipped, as the thread is not about airline safety.

 

[Politesse]

What "group identity" is the left "beholden to"?

Blacks (vs. whites), women (vs. men), trans (vs. cis), Muslims (vs. Christians and Jews)

That's why no white person or man was allowed to be in the running for Biden's SCOTUS pick or running mate. That's why DNC decided to put SC on the pole position of the primary calendar even though the primary electorate is completely unrepresentative.

So, not anything remotely like the blatant White nationalism of the fascist set?

 

[Patooka]

I could go to work, I could take a hard shit in my late 40s and see blood in the bowel and think, "Nope. Normal" I could get into a fight and think, "wait. Stop. Am I the arsehole?"

But I can always rely and depend upon derec saying middle aged white men are the real victims.

THat is my TRUE NORTH.

 

[scombrid]

As a hiring manager I’ve seen the kids with all the things versus kids that had to scratch to achieve. Man you see that resume that is polished as hell from prep school to Dartmouth and they’ve got the internships, the study abroad, the test scores. I had one that had a bunch of horticulture patents coming out of grad school. Wow, sign them up! Then put them into real work and they are just mediocre as hell. Patent kid’s dad was a professor in the ag department and kid got patents from working in daddy’s lab.

Then there’s the kid that worked landscaping, retail, military, community college and eventually got a research assistantship somewhere.

Some of the kids that come up with all the opportunities truly take advantage of them and have talent and are awesome. My classmate from high school that is a world leading research in cognitive development neuroscience is one. Granted she still had to break a gender barrier and outperform males in undergrad and early grad school before she got her Cambridge and Chapel Hill credentials.

A lot of the kids that have all of the crap handed to them turn out to be Jared Kushners. It’s why certain athletics like soccer are so fucked in this country. They’re dominated by mediocre athletes that come from money with the majority priced out of development leagues. It’s why tiny Norway fields strong teams in soccer, xc skiing, kayak sprint and marathon and the US perennially fails save a rare cases. The system is rigged so that the people that start ahead stay ahead irrespective of personal merits in sports, academics.

AA/DEI is meant to broaden the search so that merit truly is the measure.

The kid from the wrong side of the tracks that has a slightly lower test score is likely smarter than the kid with all the prep school and tutoring.

We really should look past the first polish unless we like being in a world run by Trumps, Kushners, execs at 3M that hid what they knew about PFAS for decades…