I recently got my PhD in physics at University of Arizona but found a job as an actuary where I have to learn rigorous statistics for the first time. I found this video very interesting, and I enjoyed watching it, but I have some probably common pushback.
The professor mentions f(everything else) to include all academic feedback. I was a bit suspicious of this at the time because how is it possible to hold everything else constant? Turns out the suspicion was correct, as when the students ran the regression, they only included GPA and MCAT scores. There is nothing wrong with the regression the students ran, but it's not the same as just saying f(everything else).
There could have been other predictors within (everything else) besides GPA and MCAT that could have had multi-collinearity with race and gender. Was this accounted for?
My PhD first supervisor is very good at stats. The PhD project that I handled also involved a lot of different possible parameters and explore correlation among themselves based on both fixed and random effects. Totally agreed with what you suggested, is that there could be other common variables (not fixed / constant) that worth to be explored and add in the stat. run despite gender and race, for instances, income, demography, number of family members, etc. I was working in a renowned University in Asia before, the Uni had a scoring system to run the stats for students when allocating on campus dormitory. Some common parameters they used, included distance (between students' home and school hall), academic merits (e.g. GPA), number of school clubs joined, number of voluntary services participated, and so on.
Back to this video, even though the other variables may seem to be less considered (due to course nature to make the stat project more concise and easier to understand and present, etc.), who knows after running the stats, there may be other interesting or even surprising results to see.
Jumping the gun at the end. #5 is not fleshed out -- probably can't be fleshed out. You can't predict from the data the competence of medical school students once they've graduated and in their ensuing careers. You've only just contributed to the thesis that schools have so far selected applicants based on race rather than on those other criteria.
Most people would rather have a doctor who graduated near the top of their class (great grades and recs leading to a good residency) than a doctor who graduated near the bottom. How else should competency be predicted? Maybe you'd be happy pulling a name out of your hat? Professional schools (medicine, law, engineering) should be 100% color and gender blind. If we end up with a professional class of Asians, good for them, and good for us.
You can see how the students are testing on their USMLE exams. At UCLA they had a very high failure rate last year. Historically low. But I haven’t seen the numbers broken down by race. In the past you received actual scores in the tests. It’s how you get into residency programs. It’s now pass/fail. There is data out there on pass/fail rate of the Law Bar Exam based on gender. Some states have now made the Bar pass/fail.
Understood. But if the USMLE is a filter, then medical school adcoms are doing a poor job selecting USMLE passers, not competency as doctors. This video can argue that medical schools adcoms are selecting based on race and gender, which is plain as day to see without this video, but it can't honestly speak about competency, which it hasn't even defined.
I'd like to add, in case this discussion goes no further, that dramatic as it is to focus on "the doctor" and his or her ability, it's undue and myopic. We rail against the demotion of meritocracy, but ignore declining health in our populations. We seem less interested in good outcomes for all and more interested in ensuring individual career successes, veiled as it is in moral outrage. Many more than doctors are involved in good health outcomes.
To piggy back onto this which describes acceptance rates by race into medical school, roughly 25% of available residency slots, med school grads apply for residency programs in order to continue their education, 25% are going to non-citizen graduates of foreign medical schools. And as the joke goes, what do you call the guy who graduated bottom of his class in medical school? Answer: Doctor. I’m a retired RN, lots of factors contribute to who ends up a good doctor and who is mediocre at best.
Timeout - Especially at UATX - Why omit an ensuing discussion debate. Lies damn lies and statistics. All the data analysis shows is that something occurred outside a probability, not in particularly why, ie sexist racist, incompetent etc If we use this exact methodology for nursing schools would it prove we were sexist in choosing nurses or would it just say "there seems to be a difference we would not have predicted on probability. That is where the discussion or debate begins. There also is a very strong suggestion that the previous methodology on med school selection created competence. Where is the evidence for this?.
No surprise on this, but nice to see the math. Sounds like U of Austin would be the perfect place to offer a course on the suppression, demonization and censorship of "alternative" therapies such as ivermectin, HCQ and of course, the one that knocks it out of the park on the Kory Scale, chlorine dioxide (ClO2).
Thank you. White males will have to make do in the Carribean or Phillipino schools and allow DEI to flourish. After all, its more important that your doctor show empathy than diagnostic competence or brilliance. Of course, the credentialed elite will still go to their top tier docs......
Almost certainly "devastating," but surely not “surprising." Makes me wonder if the professor has been hiding under a rock. Or maybe he's too young to have interacted with the recent decades of our best and brightest of medicine.
Pretty cool application of standard statistical analysis. This is the right way to turn the rather boring subject of statistics into a research story students will be interested in. Well done!
I recently got my PhD in physics at University of Arizona but found a job as an actuary where I have to learn rigorous statistics for the first time. I found this video very interesting, and I enjoyed watching it, but I have some probably common pushback.
The professor mentions f(everything else) to include all academic feedback. I was a bit suspicious of this at the time because how is it possible to hold everything else constant? Turns out the suspicion was correct, as when the students ran the regression, they only included GPA and MCAT scores. There is nothing wrong with the regression the students ran, but it's not the same as just saying f(everything else).
There could have been other predictors within (everything else) besides GPA and MCAT that could have had multi-collinearity with race and gender. Was this accounted for?
My PhD first supervisor is very good at stats. The PhD project that I handled also involved a lot of different possible parameters and explore correlation among themselves based on both fixed and random effects. Totally agreed with what you suggested, is that there could be other common variables (not fixed / constant) that worth to be explored and add in the stat. run despite gender and race, for instances, income, demography, number of family members, etc. I was working in a renowned University in Asia before, the Uni had a scoring system to run the stats for students when allocating on campus dormitory. Some common parameters they used, included distance (between students' home and school hall), academic merits (e.g. GPA), number of school clubs joined, number of voluntary services participated, and so on.
Back to this video, even though the other variables may seem to be less considered (due to course nature to make the stat project more concise and easier to understand and present, etc.), who knows after running the stats, there may be other interesting or even surprising results to see.
Very informative. The explanation of the model was very helpful. Keep up the great work UATX.
Jumping the gun at the end. #5 is not fleshed out -- probably can't be fleshed out. You can't predict from the data the competence of medical school students once they've graduated and in their ensuing careers. You've only just contributed to the thesis that schools have so far selected applicants based on race rather than on those other criteria.
Most people would rather have a doctor who graduated near the top of their class (great grades and recs leading to a good residency) than a doctor who graduated near the bottom. How else should competency be predicted? Maybe you'd be happy pulling a name out of your hat? Professional schools (medicine, law, engineering) should be 100% color and gender blind. If we end up with a professional class of Asians, good for them, and good for us.
You can see how the students are testing on their USMLE exams. At UCLA they had a very high failure rate last year. Historically low. But I haven’t seen the numbers broken down by race. In the past you received actual scores in the tests. It’s how you get into residency programs. It’s now pass/fail. There is data out there on pass/fail rate of the Law Bar Exam based on gender. Some states have now made the Bar pass/fail.
Understood. But if the USMLE is a filter, then medical school adcoms are doing a poor job selecting USMLE passers, not competency as doctors. This video can argue that medical schools adcoms are selecting based on race and gender, which is plain as day to see without this video, but it can't honestly speak about competency, which it hasn't even defined.
I'd like to add, in case this discussion goes no further, that dramatic as it is to focus on "the doctor" and his or her ability, it's undue and myopic. We rail against the demotion of meritocracy, but ignore declining health in our populations. We seem less interested in good outcomes for all and more interested in ensuring individual career successes, veiled as it is in moral outrage. Many more than doctors are involved in good health outcomes.
The USMLE exams are a measure of competency. They are checking your level of medical knowledge.
Hate to say we deserve what we're getting, but I think we do.
To piggy back onto this which describes acceptance rates by race into medical school, roughly 25% of available residency slots, med school grads apply for residency programs in order to continue their education, 25% are going to non-citizen graduates of foreign medical schools. And as the joke goes, what do you call the guy who graduated bottom of his class in medical school? Answer: Doctor. I’m a retired RN, lots of factors contribute to who ends up a good doctor and who is mediocre at best.
Timeout - Especially at UATX - Why omit an ensuing discussion debate. Lies damn lies and statistics. All the data analysis shows is that something occurred outside a probability, not in particularly why, ie sexist racist, incompetent etc If we use this exact methodology for nursing schools would it prove we were sexist in choosing nurses or would it just say "there seems to be a difference we would not have predicted on probability. That is where the discussion or debate begins. There also is a very strong suggestion that the previous methodology on med school selection created competence. Where is the evidence for this?.
Were stayin' in a Holiday Inn full of surgeons
I guess they meet there once a year
They exchange physician stories
And get drunk on Tuborg beer
Then they're off to catch a stripper
With their eyes glued to her G
But I don't think that I would ever let them cut on me
- Jimmy Buffet
No surprise on this, but nice to see the math. Sounds like U of Austin would be the perfect place to offer a course on the suppression, demonization and censorship of "alternative" therapies such as ivermectin, HCQ and of course, the one that knocks it out of the park on the Kory Scale, chlorine dioxide (ClO2).
we are happy to assist in such an endeavor -
Michelle Herman
CEO, Snoot! Spray
www.snootspray.com
www.TheUniversalAntidote.com
Thank you. White males will have to make do in the Carribean or Phillipino schools and allow DEI to flourish. After all, its more important that your doctor show empathy than diagnostic competence or brilliance. Of course, the credentialed elite will still go to their top tier docs......
White men don’t have empathy? And all people of color have empathy? Pretty bold and racist statement.
I WAS BEING VERY VERY SARCASTIC.......and thank you....
White people aren't responsible for the lack of intelligence of any femographic.
Almost certainly "devastating," but surely not “surprising." Makes me wonder if the professor has been hiding under a rock. Or maybe he's too young to have interacted with the recent decades of our best and brightest of medicine.
Pretty cool application of standard statistical analysis. This is the right way to turn the rather boring subject of statistics into a research story students will be interested in. Well done!