MCATs, Affirmative Action, and DEI, Oh My

[Bomb#20]

Both -- it's a hypothetical specific example of what's implied by the statistical facts. A Native American with a 509 MCAT is 1.0 standard deviation above the average admitted Native American, so he's very likely to be accepted. An Asian with a 509 MCAT is 0.9 standard deviations below the average admitted Asian, so he's very likely to be rejected.

Only if you assume Normal distributions with no skew and you also assume different thresholds on the basis of race.

I don't have any information right now that leads me to conclude that both of those are factual. I do, however, understand the risk and the perception of the latter.

I assumed monotonicity and I assumed the means and SDs in this table. If you know of a distribution with those stats that fails to exhibit the property I described, share.

(I saw your earlier exchange with Derec; to construct your counterexample you needed to have one ethnicity's SD much larger than the other's. The two SDs in the table are 6.8 and 6.0.)

 

[Jimmy Higgins]

The notion that an Asian with an MCAT of 509 who gets turned away for being Asian from the same school that delightedly snaps up a Native American with a 509 MCAT has received "fair treatment", and is "represented" because the school accepted some other Asian with a 515 MCAT, is absurd on its face. The guys who got in represent no one but themselves; and claiming the guy who got rejected on account of his race was treated fairly because of what color some other guy is is an utter betrayal of the whole concept of fairness.

Is your example fact based or simply hypothetical?

Both -- it's a hypothetical specific example of what's implied by the statistical facts. A Native American with a 509 MCAT is 1.0 standard deviation above the average admitted Native American, so he's very likely to be accepted. An Asian with a 509 MCAT is 0.9 standard deviations below the average admitted Asian, so he's very likely to be rejected.

What about the Asian with 505 on the MCAT that worked while going to college to help support his mother and brother getting in over the Asian with a 511 that didn't have to manage any level of adversity?

For people that aren't making this about race, all you are talking about is race.

 

[Jimmy Higgins]

Make the case that MCAT plus GPA should be the only things that count.

I'm not going to go around this loop with you yet again. Stop strawmanning infidels.

Strawman'ing? The entire thread has generally focused on four metrics for medical school admission.

1. Undergrad GPA
   1. Is this a good point in time to note the variability in majors to begin with?
2. MCAT
3. Medical School acceptance rates
4. RACE

We know medical schools are having to choose between a lot of very good candidates. Candidates have to include:

• Transcript
• Standardized Test Scores
• Resume including
  • Internships
  • Work
  • Community Service
  • Other Extra-curricular activities
• Recommendation Letters
• Essay?
• Interview(s)
• Personal History

Job interviews often don't care as much about metrics (if at all), as much as personality, compatibility, and competence. Interviewers are often measuring intangibles, signs that they look for in a good match for employment. The graduation rates for medical students in spite of the MCAT is incredible. While some have spun 92% graduation rate for a relatively meager MCAT score as awful, 92% is empirically high, all things considered. Which implies the medical schools are looking for and finding the right attributes for medical school students even if the MCAT and/or GPA wasn't as shiny.

So in conclusion:

• Asians are admitted at near the highest rate of any class of people (and represent a much high percentage (300% or greater) of students relative to their percentage of population)
• There is a clear relation of higher scores means higher acceptance rates
• An insanely high super-majority of accepted medical students graduate
• We know for a fact we have a terribly incomplete data set regarding Med School applicants to even consider the ability to compare them

I can't even see what the problem is, let alone evidence to support it being a "problem".