“It is not clear how employers would be informed of the narrative-based performance of students in school.”
Is an employer demanding to see your GPA such a common thing? I’ve seen it used as a weed out for things like internships but even they will bend the rules if you actually talk to someone behind the curtain.
I graduated in the aughts, but at that point every single employer asked for the GPA if you were straight out of school.
Most places it was just a filter for interviews, so you could back-door in, but in other places it would gate hiring even if you got an interview some other way. One place I got an interview because I knew someone, and after a successful interview, they rejected me once they asked my GPA (their HR form required it to be filled in for anyone coming straight out of college).
In the 90s I found not only would they want to see GPA but often the full transcript. In more than one interview I had people commenting on grades I had for specific classes.
That place that rejected me on GPA post-interview actually asked for my transcript after hearing my GPA to see if they could justify hiring me based on which classes I passed or failed. I was a rather equal-opportunity poor student, so that was not a success.
My recollection is that asking for undergrad GPA used to be pretty common--and maybe even transcript (though I don't really remember and certainly don't remember discussions about specific class grades). Though I got a Master's directly after undergrad so I may not have had the standard post-undergrad interview experience.
It was very common for on campus recruiting. And, really, there it makes some sense? Even if you disagree with the idea that people can be ranked, at large, if everyone is aware of the ranking and working with it, you are almost certainly best served getting the people that are excelling at what they are trying to get ranked at.
As soon as you open up your search to things across rankings, or in areas without rankings, though, I can't but imagine the utility of it drops off?
Oh, for sure. You should continue to do extra searching at that level. But I think it is mostly fair to assume that the high end of one ranking is a safer bet on finding talent than any other location.
It feels like the "search where the lamp post is," but I think it is different. It is close to the "we have data, it must be useful."
And there are plenty of jobs (see IB, consulting, etc. mentioned) that have an endless appetite for young, hard-working bodies that you only really expect some percentage of to work out anyway. If you miss some diamonds in the rough or late bloomers, they probably wouldn't have fit in anyway and you're fine with your simpler heuristic.
It is common for employers to use GPA to screen candidates for intern and entry level positions. As a practical matter employers receive so many applications that they just don't have the resources to interview everyone, so they need a quick way of filtering down to a reasonable number. Using a GPA cutoff is unfair and leads to the unintended consequence of students and schools both optimizing for that metric instead of for real learning. But at the same time there doesn't seem to be a good alternative that works better at scale.
How does the student performance compare to other institutions? If most students learn the material, then they deserve a good mark.
Having taught three very different classes, I can tell you there is a bell curve to the per class bell curve. Undergrad intro was a bell curve, CS Grad remediation was tragic, mixed ML was mostly A's, they were very engaged students
I mean curves apply artificial scaling to data of course how one decides to apply them. In some contexts it means that there's always someone failing because in the pool of data you're fitting to, they performed the worst (relatively speaking) but perhaps only in that pool.
Overall the worst performer could be excellent in a bigger/wider context. There's no standard "curving" procedure which leaves a lot of ambiguity and potential variation in the results. Often someone just means fitting it to a normal distribution and scaling accordingly. Not everything makes sense in a normal distribution even though people like to default to it in the absence of a known distribution.
This gets even more complicated when you start talking about arbitrarily curving across different work, perhaps different courses, across different semesters and so on. Overall you may end up with some weird weighted distribution of your actual performance that's a bit nonsensical. I'm not here to say tests are great metrics but consistency in metrics is usually a good thing or at least a consistent process for attaining them.
Where and when I went to school, there was also a bell-curve to the classes that have a natural semester for students to take them (e.g. if most students take a class in the Fall, the Spring semester class will tend to have poorer students).
The article linked doesn't add much but if you read the original report[1] it shows the percent of A's going from 60% in 2011 to 70% in 2019 and then 79% in 21. The report is about the rate of change, not so much the actual value.
The original article is more interesting as well in that it talks about the anxiety that grades bring and if abolishing grades altogether something to consider.
> If most students learn the material, then they deserve a good mark.
Yes, if you're trying to prepare your students for applying for jobs that will have GPA cutoffs, this makes sense. But if you're trying to help outsiders (and insiders, as the article notes — like award committees) discern which students are relatively stronger, this system complicates things.
Most elite universities have been forced (by market forces) into doing what Harvard is doing. The proportion of firsts being awarded by Cambridge, for example, had already inflated substantially 20 years ago.
The solution involves having, relatively speaking, "low" absolute grade boundaries (first-class honours at Cambridge is usually set around 70% of the marks in most subjects, third-class honours is 50%, pass without honors is even lower) on genuinely challenging exams, then using the raw grades for things like PhD admissions and scholarships. The notorious Maths Tripos is an extreme example of that strategy.
If 80% of students get A, it's very weak motivator if you are top 10% student. Essentially there is no point in working harder than some arbitrary pretty low threshold.
Most subjects has unlimited depth so it should not be too hard to design test with reasonable questions where no student get everything right and you can do normal distribution for grades (something like 5% A+, 10% A, 40% B, 30% C, 0-15% F)
This isn't surprising. The important selection filter is entirely in getting into Harvard. It's an excellent signal for some critical amount of intelligence, conscientiousness, and conformity, where each one can be traded off for some of the others - but only up to a point.
What would be really interesting is if the same effects happen at top universities in the East Asian countries. I would be fascinated to know why the divergence if we found that only 20% of students at, say, Seoul National University got top marks in their classes.
Normally the classes have harder tests at more prestigious schools, ensuring that most people doesn't reach an A. I've worked with people from Harvard, they weren't smart enough to consistently get A's in hard classes, so it seems to me like Harvard doesn't grade as harshly or have easier classes than most other prestigious universities.
I understand why they do it, they want their students to not have problems due to low grades, but that also means that a Harvard degree isn't as useful as a sorting mechanism because I can't tell the difference between top students and the rest.
> I understand why they do it, they want their students to not have problems due to low grades, but that also means that a Harvard degree isn't as useful as a sorting mechanism because I can't tell the difference between top students and the rest.
They got into Harvard and were subsequently successful there.
Do you really benefit from knowing which of two Harvard grads got a few extra questions right in a handful of exams?
> Do you really benefit from knowing which of two Harvard grads got a few extra questions right in a handful of exams?
It isn't just "a few extra questions", when professors makes tests harder they add more interesting questions that requires you to think further rather than just plug-n-chug. The difference between a 4.0 student and a 3.0 student at lets say Caltech is enormous, so yeah I care.
If you wanna make questions challenge students to think harder, by all means do so.
If you do, and an entire classroom of kids manages to answer them correctly, "I guess I have to still give some of you Bs and Cs" isn't a great response.
But that is how it works at other prestigious universities, basically every professor thinks that students doesn't learn stuff well enough and would want to test even more thorough understanding than they already do. It isn't about grading on a curve, it is about giving a good test for the students.
It isn't like students at any institution understands the material as well as the professor even when they get the highest grade in that class, but that is the goal. The better the students the closer you can get, Harvard just refuses to test that properly, or if they do test it they don't record that grade.
Edit: What I meant is that they typically don't grade on a curve, so it works as the poster I replied to wanted it to work. Realized "that is how it works at X" could be taken as I mean they should grade on a curve.
Why would you want "interesting problems that require you to think further" to be on exams?! That should be the entire point of lectures, office-hours, group problem solving, etc...
Because they test if students understand the material well enough to apply it to nonstandard problems. A person who can solve nonstandard problems is much more valuable than a person who can just solve standard problems.
Curious. I always assumed the primary purpose of an exam was to demonstrate proficiency, not mastery -- especially for undergraduate coursework. If you want mastery, you look at other work product: Projects, research, letters of recommendation, etc.
Pedantry and intellectual hazing seem endemic in academia -- from the perspective of someone who spent entirely too many years in the sector (all the way through postdoc).
Practicing to solve tons of standard problems is a waste of time, it doesn't help you grow much, which is why having interesting problems on exams is way better since it gives students a better goal to aim towards. It also gives a fairer view of each students understanding, since students tend to forget everything they crammed to pass the standard questions.
> Pedantry and intellectual hazing seem endemic in academia
I wouldn't call interesting test questions pedantry or intellectual hazing. Giving tons of basic easy questions is pedantry and intellectual hazing to me, I'd rather just have the harder questions.
Exams designed to have a 60% median score with subsequent curve adjustment are (a) wildly common in elite colleges, and (b) certainly classify as hazing in my book.
For people pursuing personal growth rather than a starry transcript, challenging tests are more like "lifting to failure" or "grinding the boss on hard mode" than hazing.
But on the other hand, yeah, for fragile egos and given the very high stakes associated with one's transcript, not everyone is there to get good and it can absolutely be devastating to never ace tests like you did in earlier years.
Different programs/eductators may have more sympathy for one group than the other, but there's no viciousness involved.
This is pretty common in education, at least in the teacher circles I've known. I know in Germany, they talk about the "Anforderungsbereiche", or requirement levels, where the first level is being able to repeat facts, do calculations, etc; the second level is understanding when you apply those facts or formulas; and the third level is understanding why those formulas work, or being able to escalate why some knowledge is relevant to a subject.
I would understand the third level (possibly even the second level) as including that ability to apply knowledge in nonstandard contexts that the previous poster was talking about - being able to see not just how to perform a calculation, and how to use it in a given context, but also how to create that context when necessary - seeing that the question on how humans move in groups could be modelled as a fluid dynamics problem, for example.
The general wisdom amongst teachers is that most exams should be able to differentiate between these three levels of understanding, so you want questions where the student can just follow the line and do the right thing, but also questions where they're being asked to apply knowledge in different contexts, prove why something is true, or explain why some fact is valuable or relevant in a given context. This isn't about grading on a curve, it's about differentiating between how far different students have come in their understanding. (And the corollary if that differentiation is that if all the students have the same understanding, it's likely that the course wasn't challenging enough, and many of the more advanced students could have been pushed more.)
That said, these regions are more about assessment in general than specific written exams, and the third level will be easier to test in some exam formats than others - a multiple choice exam leaves little opportunity to demonstrate that understanding, whereas an oral exam allows the examiner to explore the subject a bit more deeply. I studied in the UK where a single end-of-unit written exam is mostly the norm, and so these exams had to test all the different levels of knowledge, but if you're gathering results over the course of a year or semester, that might look differently.
Harvard admits quite a few people who would normally not qualify as the cream of the crop. Above average, even good - but the worst Harvard admit is probably on par with the smartest kid at the local community college. They might bring other things to campus! But that doesn't mean you want them handling your legal defense or managing your money.
> and were subsequently successful there.
If 80% of the grades at Harvard are As, this just means you're not in the bottom 20%. I would also generally assume that this 80% is not uniform - some classes or even majors will give out higher percentages of As than others.
There is also a strong selection for students who will fight to be the best, especially in ways that are easily quantified.
Students like this aren't always shy about demanding a formulaic way to get an A. You can try to fight it, grade on a curve, etc, but there's only so much time in a day and in the end it's much easier to come up with a formula that they can follow, and to let them follow it.
> It's an excellent signal for some critical amount of intelligence, conscientiousness, and conformity, where each one can be traded off for some of the others - but only up to a point.
No amount of conformity and intelligence can explain 80% A's if there is a curve that mandates otherwise. This grading distribution can only occur if the faculty permit it to, both at an individual level and at a policy level.
> No amount of conformity and intelligence can explain 80% A's if there is a curve that mandates otherwise.
A curve requiring an entire class of people who all got 90% of the questions right on an exam to have a particular distribution of As, Bs, Cs, Ds, and Fs would be insane. Curves are supposed to account for "this test is quite difficult and has an expected range of passing scores that doesn't quite fit in the ten point scale" scenarios.
If you can get into Harvard, you're supposed to be an A student.
EDIT: The parent comment makes it sound like admittance to Harvard is prima facia proof that students are already excellent. If that is the case, then why grade students in Harvard? Students at Harvard should be trusted to learn everything they need to learn.
EDIT2: Given that 80% of grades at Harvard are 'A', it seems like Harvard administration agrees that grading is not needed.
Because it indicates "this student has a good grasp of subject X".
If everyone at Harvard exits with a good grasp of Calculus, that's useful information. If only half of people from BU graduate with a good grasp of Calculus, that's also good information. BU and Harvard don't need to have the same proportions of A students for it to be a useful data point.
I think the root problem here is that some people see grades as an absolute measurement of whether someone has mastered the material, and other people see grades as a way to rank students relative to each other. If you believe the first, then it should be possible and expected for a class to be able to get all A's. If you believe the second, then it makes sense to widen the dynamic range such that you get a clear rank ordering.
To catch people who don't meet expectation. Tests are written such that A means "you performed at the level that was expected of you" and B-D are varying magnitudes of failing to do that.
This is just semantics. When tests are written to where the expected grade is an A then less than that is coming in under expectation. Also I don't know any school that lets you move to the next course if you got a D. It's enough to say you passed but not continue.
It isn’t a zero sum game. It’s possible for all students to get As and for that to still be a meaningful measurement of skill.
Grades don’t show your ability relative to others, they show your mastery of a set of knowledge and skills. You mastering something doesn’t prevent your neighbor from also mastering it.
In my day, Rank-Based grading was the system, but not just among the students in a specific class -- it was based on historical numbers. The same course was taught for many years so it was easy to get that longitudinal view.
Especially in secondary and high school, a 'C' meant you were average for the students who took the class. The tests had questions of differing levels of difficulty. If you couldn't answer the difficult questions, you clearly didn't "master" the material and therefore couldn't get an 'A'.
People with average ability would be expected to get an average on the test, so a 'C' would still mean you learned the material adequately and could move on to the next class / grade.
Feedback to the student about their level of mastery of the coursework (relative to the coursework itself, not the rest of the class).
Of course, that doesn’t have to be a letter; percentages along with specific corrections would do as well. But once there’s a convention people often use it.
Grading students relative to each other gives different information and probably not precise across cohorts at that.
The obvious solution in that case seems to be to make the test more difficult . If everyone is getting 90% right you can definitely spice things up a bit
Why? The exam is intended to make sure you learned the material. If everyone did so, the response shouldn't be "make the grades shittier", it should be "applaud the instructor for their good teaching".
My AP Physics class had 20 kids in it, 10 of us got 5s and 10 got 4s on the AP Physics exam. Should some of us had our scores lowered for a better bell curve? No; he was just a really great, engaging teacher who pushed us to achieve.
> 10 of us got 5s and 10 got 4s on the AP Physics exam
There is no problem if a subset of the total population does very well (or very badly). But if half of all AP takers got 5s and the other half got 4s, then it would be time to reexamine the test.
Similarly, if there was a study group in a class that did very well, that's no problem. But if nearly the entire class gets As, that signals that the class could be more rigorous, cover more material, etc.
I get why they don't do this — they don't want to disadvantage students as they go looking for jobs. But some schools are happy to have a reputation for for tough grading. My college had freshman courses where only 10% could get an A or A- (this amounted to 3 students, since the classes were capped at 30). It's a very selective college, so pretty much all of the kids were "A students" coming in. Among graduate schools and top employers, it's known to be a school that is very academically challenging.
> But if half of all AP takers got 5s and the other half got 4s, then it would be time to reexamine the test.
Maybe, maybe not. If you spot check and a statistically significant sample shows kids are indeed learning first year college physics effectively in AP courses, perhaps you do a little happy dance and be glad something's working.
Employers don't seem to be hiring Harvard grads and going "holy shit this person doesn't know anything they're supposed to know" to half of them. Employers seem to be generally happy what an A student from Harvard comes to them with.
> But if nearly the entire class gets As, that signals that the class could be more rigorous, cover more material, etc.
Or that you've got a self-selecting population of high-achievers.
> My college had freshman courses where only 10% could get an A or A- (this amounted to 3 students, since the classes were capped at 30).
That's depressing. "You're an A student, but I'm required to give you a B for organizational reasons" is an absoutely nasty thing to do to a college kid.
> Employers don't seem to be hiring Harvard grads and going "holy shit this person doesn't know anything they're supposed to know" to half of them.
That's because they (1) know there's rampant grade inflation and (2) use other indicators to assess candidates. And I've definitely heard people grouse about Ivy League grads, and how they'd rather hire from other places.
> That's depressing. "You're an A student, but I'm required to give you a B for organizational reasons" is an absoutely nasty thing to do to a college kid.
"Organizational reasons"? No. It's because they view grades as a tool to let kids know how they are doing compared with their peers. If you don't want to go to a school like that, there are plenty of grade-inflationary schools out there for you to go to.
The point of tests is to determine if the student learned what they're supposed to learn. If everyone gets 90% or better on a test that reflects what you're supposed to learn then it should be celebrated as a teaching success. It is not a problem.
A perfect educational system would have every student getting 100% on every test, right? That's the ideal. Unless, of course, that the purpose of the educational system isn't actually to educate people but instead to weed out certain classes of people.
"We want the best" is what every employer will always say but the truth is that they really only need someone who can do the job. We as a society put way too much importance and faith in credentials.
> If everyone gets 90% or better on a test that reflects what you're supposed to learn then it should be celebrated as a teaching success. It is not a problem.
Then perhaps there should have been more available to be learned. If most people are above 90%, then you're holding the best back. (If you're saying that more should be taught, but not be on the test... Why?)
The only way for everyone to reach full understanding at the same time is for everyone to be the same or for some people to have been held back. I view the latter as being far more achievable than the former. I would also say that holding the best back... is probably not what we want the #1 ranked University in the world to be doing.
> A perfect educational system would have every student getting 100% on every test, right? That's the ideal. Unless, of course, that the purpose of the educational system isn't actually to educate people but instead to weed out certain classes of people.
No, a perfect educational system would not have every student getting 100% on every test because not every student is the same and also such a test fundamentally provides no information to the tester.
A perfect system would probably have every student on an individual (or near-individual) learning schedule, and even there I would expect perfectly calibrated tests to average around 50% with adjustments as required for psychological impact or to improve recall. From that, you can identify what elements need to be covered again, and what can be skipped when moving forward because they are already known.
> "We want the best" is what every employer will always say but the truth is that they really only need someone who can do the job. We as a society put way too much importance and faith in credentials.
This is probably correct for your average community college. It might be correct for a average state school.
When you get to 'literally the number one Univeristy on the planet', perhaps employers are also actually looking for the best people on the planet.
> No, a perfect educational system would not have every student getting 100% on every test because not every student is the same and also such a test fundamentally provides no information to the tester.
This is a ridiculous statement. Let's assume you have a room full of perfect students. You spent the week teaching them about subject X and came up with a test to determine if they got the gist of what you taught.
On Friday you give them that test and every student gets 100%. By your reasoning, because they all got 100% this test was a waste of time and would "provide no information to the tester."
My argument is that the test served its purpose: It demonstrated that everyone understood the subject and you can move on to the next. Could some students have absorbed far more information in that same amount of time? Probably. Is that the teacher's job though? No. Their job is to teach a certain subset of a subject and do their best to make sure every student understands it well enough that they can move on to the next.
The problem with this method is that smarter students get bored out of the minds and end up feeling like they're in prison for their academic career. However, the system itself wasn't made for them. It was designed for the average.
If you feel that grades should be used to separate or weed out students who don't perform as well you're trying to twist "education" into something that it's not: A caste system.
> If you feel that grades should be used to separate or weed out students who don't perform as well you're trying to twist "education" into something that it's not: A caste system.
I would say I am trying to return education to its original purpose, instead of the world's worst but also most common form of daycare.
And the difficulty of that subject matter should reflect that expectation of superb academic excellence. Harvard has a duty to increase the relative difficulty of their coursework UNTIL a certain percentage fails. Otherwise it's just a rich persons Lions club for angsty 20-somethings.
> Harvard has a duty to increase the relative difficulty of their coursework UNTIL a certain percentage fails.
Why? What benefit is there to the world to making students flunk out of a school when that same student would have passed with flying colors elsewhere? "Sorry, you've got a great grasp of engineering that any firm would benefit from, but you'll work at McDonalds now, because Bob got one more question right. Thanks for the $100k and four years of your life!"
The goal of education is to educate, not to flunk a certain percentage of people.
> The goal of education is to educate, not to flunk a certain percentage of people.
No. The purpose of normal college is to educate.
The purpose of Ivy league school is to separate the exquisite from the mediocre. If you aren't exquisite then why are you going to $100k college? Because your parents can afford it.
There is some low income kid who actually is exquisite who should get that seat in the exquisite class.
You don't deserve to pass just because you're not an idiot. If you are even a little bit of an idiot you should not be able to get a Harvard education. You should flunk out, lose your parents money, and try again in community College.
You are not entitled to pass because you are average and your parents are rich. College is not risk free.
Right, because nothing shitty and unforeseeable ever happens to low-income kids in the middle of a semester, because kids who definitionally have fewer resources aren't held back in any way by their environment, and because the people setting the standard are unbiased angels. Dollars to donuts you've never actually taught a class.
> You are not entitled to pass because you are average and your parents are rich.
You should be entitled to pass if you can successfully demonstrate your understanding of the subject matter in question. Proposing to curve passing grades down "UNTIL a certain percentage fails" is just... nutty.
> You are not entitled to pass because you are average and your parents are rich.
Yes, which is why grades should be based on "did you learn?", not "do the grades look like a proper bell curve?"
> There is some low income kid who actually is exquisite who should get that seat in the exquisite class.
A low-income kid is the least likely to be able to take the risk of going to a college that's intentionally trying to flunk people out even if they're competent at the subject matter.
I completely agree with you and fail to see why people think otherwise. It’s just so arbitrary and against actual learning to make a test hard to get a certain spread of scores. Depending on the material, breadth, and depth the teacher would like to assess what has been taught, the test can be challenging. The goal is to learn.
> If you can get into Harvard, you're supposed to be an A student.
This seems to presuppose that the grading system is universal across all universities - i.e. that we would expect the vast majority of Harvard students to get an A, but the vast majority of students at a middling community college to get a C or D.
Is this really the case? I'm not American, but in the British education system, this is broadly considered not to be the case - if anything, the more prestigious institutions historically have had much higher standards for each grade.
Another expectation is that students' performances at university typically cannot be easily predicted from school grades or tests, and that all universities will have a proportion of students who do very well, and a proportion of students who do not do well at all. There is no real expectation that, say, the Oxbridge universities only take students who will definitely get a first, and the other universities mop up. Oxford and Cambridge are definitely more prestigious and therefore harder to get into, but they're by no means some "sure thing" where if you've passed that hurdle, you'll be able to pass everything else.
Is grading on a curve desirable? Why not, if everyone is really good, everyone gets an A? If they all fail, they get F's. I had a friend who was a professional musician auditioning for orchestras. At that level, EVERY applicant to these positions had won every music award they were eligible for since they were a kid. He never paid a dime for his music degree because he had a full scholarship. Zero debt to worry about. Still the audition outcomes were basically random because everyone trying out was perfect. So what's wrong with that?
Just from the top of my head, SKY (SNU, Korea, Yonsei) have pretty significant grade inflation -- probably not as bad as Harvard -- and a cursory Google search supports that.
Back to US schools, I'd like to see some analysis on the correlation between the proportion of # of GE courses for a program and that program's grade inflation levels. Also, compare public vs private schools' grade inflation. Schools in general have a strong incentive to send their students to internships and post-grad programs at prestigious IB firms, consulting firms, law schools, grad schools, and etc; and I wouldn't be surprised if private schools tend to have more lenient policies in relation to grade inflation to boost that goal.
Citation Needed. I don't think anyone has ever thought "ooo, a Harvard grad, they must be really conscientious." They do, accurately, see a white Harvard grad and think "50/50 chance on whether they're brilliant or they come from money."
"The important selection filter is entirely in getting into Harvard."
You are confusing cause and effect there. By giving everybody high grades, they are reducing the ability to tell Harvard graduates apart, thus increasing the desirability of getting into Harvard and decreasing the need to put in effort after being accepted.
A great many of them aren't paying tuition at all. By this point Harvard has an endowment larger than some small nations. And yet, I would imagine a similar, perhaps even stronger, trend towards getting all As in these "free riders".
It always surprises me how some people are shocked a plot of P(student learned the class material well) is starkly different from P(student learned the class material well | student was hard-working enough to secure admission through one of the most selective* processes).
* Minus the legacies and the atheletic admissions I guess.
Following the opening link to the report it shows the year was not an anomaly in the trend, rather the last (and somewhat naturally the most extreme of the trend) in the dataset.
My brother graduated from MIT 20 years ago and I recall one summer when I stayed there one of his fraternity brothers quipping “the fastest way to get a girl in Boston is to tell her you go to Harvard, and the fastest way to lose her is to tell her you go to MIT, they don’t want nerds”. Amazing how things have changed but even then it was basically an accepted truth that Harvard massively inflated grades. At the time I was told perfect GPAs were basically unheard of at MIT (back then least).
Grade inflation is real and I think it leads to significant unintended consequences. I do find myself flip flopping however on whether there should be grades at all. If you’ve read zen and the art of motorcycle maintenance, you’ll recall there is a powerful case made to eliminate grades at the college level. I still lean toward having them currently.
Another fun anecdote about those times and MIT specifically. My brother was arguably the top student to come out of the Atlanta Public School System in 1996. This was a kid used to being the best in everything he did. He starts at MIT and takes the Calc 2 placement exam (he got a 5 of course on the AP BC test) and scores an 11 out of 100. He said the teaching assistant sat him down and told him “we are not in Kansas any more, forget everything you think you know, you are not prepared to be in this class and we will not accept your AP test results.”
Unfortunately, I recall in 1996 4 or 5 students committed suicide at MIT due to the academic rigor (which is why the first year was pass fail).
I’m not advocating for this system just sharing stories told to me.
While there are softer subject escape hatches at a school like MIT (and I've known some very smart people have taken them while remaining in technology-related areas generally), it's hard to avoid math and hard science/engineering topics in general. And there are a lot of straight A students in high school (and most people at MIT were at least close to that) who were fine with memorized plug and chug high school physics suddenly hit stuff they just can't handle and struggle to just pass the class.
In my experience, MIT is actually pretty good about wanting to help students work their way through once they've been admitted. But there's definitely a firehose aspect to a lot of it a lot of incoming students aren't prepared for--even leaving aside a lot of the distractions and other elements of suddenly being away at college.
Anecdote: I taught at an online bootcamp and I swear our best cohorts were the ones before we started grading. We told them up front: "graduating" from this place doesn't mean anything--you don't get a diploma. They all knew the real test was to succeed in an interview and get a job, and they kept that focus, and worked hard and independently.
Once we started grading, suddenly the game was to get a good grade. And students have all kinds of ways of getting good grades that don't lead to jobs.
This statistic is useless without any comparison to other schools.
Maybe this is the same as the average across all of the top 20 schools?
It also raises an interesting question: should it be harder to get an A in a class at Harvard than at e.g. a third-tier school? Because the answer to that is not obviously yes.
E.g. if you take Calculus 101 and master it to some level of proficiency that is an A- at some third-tier school, should you also get an A- in Calc 101 at Harvard, or should that merely be a C there? (And if you think it should be a C, then why? Isn't calculus just calculus? Either you master the material or you don't. Why should you be graded differently depending on where you learn it?)
Because if you think grades should reflect some kind of absolute level of proficiency, then it's not surprising at all that 80% of grades at Harvard would be in the 'A' range, considering how many straight-A applicants Harvard accepts in the first place.
Should it be harder? No, I don't think so. Ideally the classes at any two schools should be roughly on par with one another. But I would assume that disparities do exist, unfortunately.
An anecdote: my father was valedictorian of his podunk high school but nearly flunked out of his first year of college. And he's a smart guy, it was just a mixture of (1) not having study skills, since he'd never really needed to study before and (2) being a little behind on certain subjects (like math) that hadn't been taught well at his high school compared to the schools his peers attended.
In a situation like that, I'd assume that an A in a class whose teacher is just phoning it in is not equivalent to an A in a class on the same subject whose teacher is serious about both teaching and grading.
This is why standardized tests exist. They're the same test no matter where you're from.
IMHO, standardized tests should always be about mastery and scoring should reflect that rather than having a simple number like 1200 or 1600. Someone could get a 600 on their Math SAT but the parts they got wrong were related to simple fractions or time-based calculations. If the only information you have is, "600 out of 800" you don't have that information.
Furthermore they shouldn't even have scores, really. They should be pass/fail based on mastery of required subjects (whatever the test is about). Either you know quadratic equations or you don't. If you got 90% of the quadratic equations correct you know how to deal with quadratic equations... You probably just made a minor mistake somewhere or were running out of time. You still "got the gist of it".
It would make a hell of a lot more sense to me if students had to take dozens of standardized tests demonstrating mastery of all facets of the high school curriculum rather than one great big test at the end of their schooling that measures their ability to remember randomly-sampled fractions of that knowledge.
>It would make a hell of a lot more sense to me if students had to take dozens of standardized tests demonstrating mastery of all facets of the high school curriculum rather than one great big test at the end of their schooling that measures their ability to remember randomly-sampled fractions of that knowledge.
Wait, if the one big test at the end was randomly sampled well, it would still provide results almost as good as requiring 10 to 30 times the same amount of test time and energy per student. I don't think this cost to benefit ratio makes sense.
>Ideally the classes at any two schools should be roughly on par with one another. But I would assume that disparities do exist, unfortunately.
Do you really think an intro to calculus class at the local community college and at Stanford should be the same class? Because I guarantee that the standardized class will either bewilder most of the community college students or bore/waste the time of many of the Stanford students.
If the grades are supposed reflect profficiency then sure. The program should be the same.
But one should argue that those good schools should have a more difficult program.
This of course leads to the problem known in Poland (before the 2009 reform): the top schools ha# harder programs. For example if you studied French, all the students allready knew it before joining the university. They learn the advanced topics.
Or if you studied maths on the entrance exam they would ask you about topics that are not part of the highschool curriculum.
They changed it recently, but Im not convinced that for the better.
Now everyone writes the same final exams, each university doesnt run their own exams anymore - at least in theory. Some still do
When I was a grad student (back in the Old Stone Age), the minimum passing grade in graduate courses was an A. (This was based upon the theory that we were cherry-picked from the set of applicants.) This may be Harvard's premise (the people I've known who went to Harvard were generally outstanding), or maybe not. I would be interested in seeing the correlation between legacy students (the ones who got in because a parent attended) and grades.
In 30 years of teaching first- and second-year computer science courses, I always saw a bimodal distribution, with one mode around a C and the second at a B+/A- level. (In first year, it often was trimodal, with the third mode at an F-, students who gave up early in the course and didn't withdraw). I would be astonished if there were not something like that going on in Harvard's first year CS.
Mr Turley quotes William F Buckley on preferring the first 400 names in the Boston phone book to Harvard faculty. I would prefer to listen to a conspiracy theorist explaining why the lepton samurai of Jupiter were about to invade the earth than to listen to Mr Buckley, even if he were still alive.
I took CS grad courses at Stevens Tech in Hoboken, NJ during my undergrad. The minimum passing grade to stay in the grad program was a B. Tons of curves were had.
I don't know anyone in tech who is impressed with traditional IVY schools. It's not a negative signal, but I don't really think they have rigorous CS or EE programs.
MIT, Stanford, and CMU do stick out to me as a positive signal but I could care less about anything else, including Harvard.
The blog post seems to say that the Harvard grads couldnt be squeezed out like a lemon with 100h weeks ("balance between hunger and burn out")...
As much as I dont like grads from top schools. Here they showed some actual smarts. You dont need to work 100 hour weeks in a sweatshop that tries to "balance you between burned out and hungry". Probably that grad has rich parents too, so they can switch jobs to a boring corporate job. They have a safety net not to be exploited with 100hour per week gruel. Come on? Is that some success story? 100 hour weeks is eithet a sweatshop or shitty work organization.
I bet that lady wanted that motived, hard working harvard grad for 15 dollars per hour and also with 50 hours of UNPAID overtime. That's jusr the typical sweatshop approach. Also they should "think big" but not too big like for example "we should pay better in our sweatshop".
I looked who she was - and seems her own partners kicked her out from that company that she was supposedly building by exploiting people.
To get to Harvard, you need to be a valedictorian or close to that. And not every valedictorian gets to go to to Harvard. It’s not a surprise then that Harvard is full of overachievers with straight A’s.
There's been a general trend of grade inflation in U.S. universities, accelerating in the 1960s (during the Vietnam war), and then again in the 1990s [0]. In some sense, this might be because the role of the student shifted more towards that of a consumer purchasing a product.
Harvard average GPA was a 3.0 in 1967, now it's a 3.45 [1].
There are some universities that have tried to actively prevent inflation (e.g. Purdue with a 2.73 average in 1986 to 3.09 in 2012 [2]).
God I hate this take. This mythologizes grades as an end rather than a means. I much prefer the world where tests are written with the expectation that people who've mastered the material get an A rather than the endless hamster wheel where the test gets harder until you get a distribution centered at C.
I would have burnt out on my schooling so fast as to make it worthless. How demoralizing it must be when every test is designed so that you won't know all the answers. Why even bother trying? I'm not in school to compete with other students, I'm there to learn.
I'm near finishing my undergrad at Harvard extension school. Many of the courses (especially in CS) are taught alongside traditional Harvard students. Same assignments / tests / lectures, they just have to be physically in the lecture hall.
The work is rigorous, and the grading is usually harder than the state school I attended long ago. (I have taken a few non-math / non-CS courses that were easy A's).
Maybe grades have inflated, but overall the students are really good. Grading really harshly just to "separate the top" or whatever doesn't make sense. They all mastered the material in "Introduction to Computer Science" in the semester.
It does make sense because grades no longer mean anything if everybody gets the same grade.
The most informative grades would be uniformly distributed, it’s an opportunity for the geniuses to shine and give a signal to profs about who to hire for their PhDs.
If PIs or employers want to discriminate based on class ranking, they can ask for it and do exactly that. Why conflate that with grades?
A common understanding of a grade is as an indication that the course material has been mastered. Higher grade = more mastery. This is a measure independent of the performance of fellow students.
Because it’s nontrivial to convert a grade into a ranking when everybody has the exact same grade.
If you gave a test out and the average was 99 percent, you would be ranking noise. They should make the course harder so even out the distribution of grades.
The most informative distribution (ie maximum entropy) of grades is uniformly distributed. The least informative is when all grades are the same.
> Grades are most useful to a professor if it indicates class ranking of who has potential to do research level work.
There's a bald assertion here that lots of those professors would stop you at; that good grades and capacity for research work are the same (or even necessarily closely correlated).
> They all mastered the material in "Introduction to Computer Science" in the semester.
There is no way all of them did. Eric Mazur tested this for people taking physics at Harvard, he found that even though every student got an A they didn't actually learn the material. He tested before and after on standard physics understanding, they didn't improve even though they got A in the class.
Alright, I do agree with that. Many people are incredible overtrainers on data without the ability to generalise. Not sure if that's any more true here than everywhere.
Thanks for posting that talk. Its fun to watch a lecture by someone who knows how to deliver it, even if you only have a passing interest in the subject.
i think it's fine if universities grade on whatever scale they want, or none, and it's not clear who is hurt by this.
I guess if you have to hire Harvard new grads, want to base your decision on their academic performance, and would prefer more fine-grained rankings of performance, this is not in your interest.
The very top students may also wish they had more chance to distinguish themselves. However the way it seems to work in many schools with grade inflation is that it's easy to get an A, but still pretty hard to get an A+, so there's still some opportunity here.
I'd also add that at Princeton, attempts to combat grade inflation were very unsuccessful and resulted in strange unintended consequences, like students carefully strategizing which sections of courses to take to avoid those believed to have heavy competition.
>I'd also add that at Princeton, attempts to combat grade inflation were very unsuccessful and resulted in strange unintended consequences, like students carefully strategizing which sections of courses to take to avoid those believed to have heavy competition.
Certainly, if you're going to major in some quantitative field, you can only avoid so many of the courses in that major. But, if you're really interested in optimizing for GPA, you can often still major in some variant of the major that provides more flexibility and--where you have electives--take classes that are easy (for you).
What should the average be? A 'C'? I doubt it. Let's assume most of the people at Harvard wanted to be there. Wouldn't they work hard to get there and thus once there do their best to succeed?
If you're already in the top 10% of all students in the USA, you can still be average in comparison to your peers.
A 'C' grade at Harvard should still be pretty good in the job space because: 1) you were good enough to be admitted to Harvard, and 2) you are representative of an average student at Harvard.
Says who? I find mastery-based grading much more logical.
Take the Finnish or Estonian school system, it's very rarely curved like in the States. If you master the material you get a good grade. PISA results seem to indicate that works well.
Grade inflation is an obvious problem. It does a disservice to students, making them believe they are exceptional at something they are actually mediocre at. Some rants related to this and how awful grade inflation is for students and the world.
AP score data is especially indicative of this. In AP classes, most high school students earn As from their teachers (I estimate around ~60%). Yet, when you look at the AP scores, few get 5s on exams (the equivalent of an A). For example, 5-10% get 5's on AP English and Science-related exams. Scores are a bit higher on Social Sciences (10-15% 5s) and math/CS (~25%). But only ~50% of students even get above a 3 on the exam (the equivalent of a C). So there are people who essentially get a D by the standard (a 2) that are getting an A in their high school class.
I could go on with countless examples of how students aren't nearly as capable as their grades would indicate. At McKinsey, I interviewed over 100 people from top schools with high GPAs. Many couldn't solve simple math problems when given the problem in the context of a real-world case.
At out top institutions with our top students, we should be pushing them extremely hard and measuring them against a higher bar. I find it disappointing that they are being measured against the same (or even a lower bar) than students at other institutions. It's so bad that it seems clear that a Harvard Education isn't any better than a Penn State, Nebraska, or even UC Riverside education. The only difference is having the Harvard brand and network. It's an embarrassment. And it's not just a problem at Harvard. It's a problem at every top institution.
Grade-motivated students don't put in the extra work when little effort still earns them As. One example stands out. I run an edtech company, and in our early days, we ran an experiment at doesmyessaysuck.com. You could submit your essay, get a score, and 2 pieces of feedback on how to improve the content and structure of your writing. We did about a thousand essays before abandoning it. I always think about one student. He submitted an essay we generously gave a C (it was really bad). It was incoherent, poorly structure, didn't answer the prompt, and lacked sound logical reasoning. He responded we were wrong because he had gotten a 93% on it.
Also – https://www.gradeinflation.com/ is a good read. It's out-of-date data. But it proves the point even further than this article.
Note: There is some research out there that claims grade inflation is false and that more students actually have learned more than in previous years (i.e., the academic bar hasn't changed, but more students are above it). However, this research is unconvincing (and ripe with errors). There's essentially been zero improvement (and even a decline) in basic math and literacy skills over the past couple of decades (as measured by standardized tests). Yet grades are much higher. And high school and college graduation rates are up by about 10 percentage points since the early 2000s. As it turns out, when your only measure of success is graduation rate, you end up with more graduates – even if those graduates don't have any skills.
> In AP classes, most high school students earn As from their teachers (I estimate around ~60%). Yet, when you look at the AP scores, few get 5s on exams (the equivalent of an A).
There's absolutely no problem with this.
Your grade in your AP class is a measurement of your achievement at the high school level. An A indicates you did what's expected of you as a junior/senior in high school.
The AP exam is measuring whether you achieve at a first year college level enough to skip that first year in the subject.
More people should get As in AP Physics than get 5s. If they measured the same thing you wouldn't need the AP exam.
This is false. AP classes are meant to be college-level classes. The goals of these classes is to earn scores that yield college credits. The tests exist to provide some form of standardization around understanding a student's performance against a standard in these classes. AP exams are not measured on a curve like an SAT (although even the SAT has massive score inflation. AP exams are measured against a standard for learning – and the standards are clearly not being met as clearly evidenced by the scores. Similarly grades are meant to be a measurement of learning against a standard. Yet, grades have increased and learning outcomes have not increased. This is a fact.
> AP classes are meant to be college-level classes.
They are high school classes at an honors level that may generate college credit. A two on the exam is deemed “possibly qualified” to skip a year of college, for example, but colleges won’t take it. Many won’t take the three score’s “qualified” either.
For purposes of your GPA and high school physics, that’s not a D.
The AP physics exam gives you a 5 if you miss like half the questions, too. That doesn’t fit a ten point GPA scale very well.
Why does this even matter? In the end as long as you pass the required courses, you'll get a degree. Who checks grades? Most don't even check that a claim of degree is true. The only loss would be that of the student not actually the learning material due to lenient grading.
Q: What do you call a doctor who graduated at the bottom of their class? A: "Doctor"
I think the point is that there could be some meddling to make sure people pass. I am not purporting that's the case I just think that's why it's mentioned. I went to a school that was nowhere near Harvard level and saw it regularly, people sleeping in class or who didn't do a single assignment who still graduated with me. I think it was a money thing. Still not saying that's happening at Harvard just that it wouldn't be the most shocking thing.
That said, I agree with other commenters that you set this situation up with the rigorous acceptance criteria. You're testing the people who got into one of the best schools so presumably if your selection criteria works they should all excel at the courses.
Probably a positive trend if college students end up worrying less about their GPA and spending more time learning things they are interested in / trying out different careers / making friends and memories.
From my experience, college students worry enough about grades without forcing the curve to be harsher than it currently is.
Interestingly this is also a problem at the (expensive) private schools that feed students into these elite colleges.
You can give out super hard exams and make them sweat/cry/worry, but you'll be given tons of shit if you actually fail them, especially if their parents are large donors. Only exceptions here are for students that failed so badly they're no way they can pass, even if you apply a crazy curve on top.
now many (most?) of those students are extremely talented and deserve the grades they get, but it's still a bullshit system that encourages failing upwards.
This is my personal experience with the grading systems and how I find (personal opinion) assigning grades as a very hard problem.
I did my undergrad outside the US but in a university that follows American system. My department culture was that you don't expect curves and grades are based on what you get. So if we are all getting F then that is F for all people. No one passes a course just for the sake of passing. While these extreme cases never happened (at least when I was there). if it happens then this tells about quality of the course itself not only students effort and performance.
I took a general relativity class during my undergrad. I like this field and topic. And while I am currently doing particle physics but I always liked GR. I was doing well regarding the topics but never got good grades. I struggled with its exams in particular with this particular professor. I got C although I managed not only to have a good grasp of the subject but went so far as to work on specific problem and expanded it into a paper, participated in a conference and got it published later into a peer-reviewed Journal.
While when I moved for a graduate school is the US. Things is very different. Many of the undergraduate I teach were always expecting high grades anyway. curving is very much the norm and the negotiations is about how much to curve. The normal distribution of the grades is a religious belief to some people in the universities administration. And I understand it is hard to assist professors teaching and some use this as an evaluation metric but it I feel like it doesn't addressa core thing. Some people will pass even with high grades and they did not get the core of the subject. This is not about they deserve more than me or not. There is a fundamental question of are we grading on effort to learn or actually what people learned. The grading criteria will differ much based on our answer to this.
In your profession effort to learn new thing is the most important thing. But with the degrees that assumes that you are ready if you have it is that the expectation or your effort and ability to learn. But how can wr measure that specially on scale.
During my graduate courses I took a QFT class taught by condensed matter physicist and was focused on his area of research. I wasn't interested and it was very tough. I don't feel I understood most of the things and I struggled to the point I did not submit the final exam (take home). And to my surprise I got an A because I was the highest grade in my class (we were few anyway). I learned a lot of QFT later on my own but I am giving this example on two completely different personal experiences that hopefully deliver the idea.
It is hard to design a grading system on a scale that will try to measure different things. and the universities marketing for degrees is doing much harm ( for financial reasons). And also graduate schools and companies don't usually want or sometimes doesn't have resources to look into details about applicants institutions even if they have information about what their grading works. Hell I don't even think it is possible to have one unified system per university (Is it even good thing to do?).
90% of employees in a 10% stack ranking environment get an “A”. USMLE has a 90+ % pass rate.
1% of the population dies every year. unemployment + inflation is between 5 and 15%, a state of permanent misery index range. in the steady state, it’s enough creative destruction.
139 comments
[ 471 ms ] story [ 3838 ms ] threadIs an employer demanding to see your GPA such a common thing? I’ve seen it used as a weed out for things like internships but even they will bend the rules if you actually talk to someone behind the curtain.
Most places it was just a filter for interviews, so you could back-door in, but in other places it would gate hiring even if you got an interview some other way. One place I got an interview because I knew someone, and after a successful interview, they rejected me once they asked my GPA (their HR form required it to be filled in for anyone coming straight out of college).
As soon as you open up your search to things across rankings, or in areas without rankings, though, I can't but imagine the utility of it drops off?
It feels like the "search where the lamp post is," but I think it is different. It is close to the "we have data, it must be useful."
Having taught three very different classes, I can tell you there is a bell curve to the per class bell curve. Undergrad intro was a bell curve, CS Grad remediation was tragic, mixed ML was mostly A's, they were very engaged students
Overall the worst performer could be excellent in a bigger/wider context. There's no standard "curving" procedure which leaves a lot of ambiguity and potential variation in the results. Often someone just means fitting it to a normal distribution and scaling accordingly. Not everything makes sense in a normal distribution even though people like to default to it in the absence of a known distribution.
This gets even more complicated when you start talking about arbitrarily curving across different work, perhaps different courses, across different semesters and so on. Overall you may end up with some weird weighted distribution of your actual performance that's a bit nonsensical. I'm not here to say tests are great metrics but consistency in metrics is usually a good thing or at least a consistent process for attaining them.
The original article is more interesting as well in that it talks about the anxiety that grades bring and if abolishing grades altogether something to consider.
1: https://www.thecrimson.com/article/2023/10/5/faculty-debate-...
You've also had some long-standing anomalies like MIT using a 5.0 GPA scale rather than 4.0. (And at least used to be Pass/Fail Freshman year.)
I do get that grades aren't an unalloyed good.
Yes, if you're trying to prepare your students for applying for jobs that will have GPA cutoffs, this makes sense. But if you're trying to help outsiders (and insiders, as the article notes — like award committees) discern which students are relatively stronger, this system complicates things.
The solution involves having, relatively speaking, "low" absolute grade boundaries (first-class honours at Cambridge is usually set around 70% of the marks in most subjects, third-class honours is 50%, pass without honors is even lower) on genuinely challenging exams, then using the raw grades for things like PhD admissions and scholarships. The notorious Maths Tripos is an extreme example of that strategy.
Most subjects has unlimited depth so it should not be too hard to design test with reasonable questions where no student get everything right and you can do normal distribution for grades (something like 5% A+, 10% A, 40% B, 30% C, 0-15% F)
What would be really interesting is if the same effects happen at top universities in the East Asian countries. I would be fascinated to know why the divergence if we found that only 20% of students at, say, Seoul National University got top marks in their classes.
I understand why they do it, they want their students to not have problems due to low grades, but that also means that a Harvard degree isn't as useful as a sorting mechanism because I can't tell the difference between top students and the rest.
They got into Harvard and were subsequently successful there.
Do you really benefit from knowing which of two Harvard grads got a few extra questions right in a handful of exams?
It isn't just "a few extra questions", when professors makes tests harder they add more interesting questions that requires you to think further rather than just plug-n-chug. The difference between a 4.0 student and a 3.0 student at lets say Caltech is enormous, so yeah I care.
If you do, and an entire classroom of kids manages to answer them correctly, "I guess I have to still give some of you Bs and Cs" isn't a great response.
It isn't like students at any institution understands the material as well as the professor even when they get the highest grade in that class, but that is the goal. The better the students the closer you can get, Harvard just refuses to test that properly, or if they do test it they don't record that grade.
Edit: What I meant is that they typically don't grade on a curve, so it works as the poster I replied to wanted it to work. Realized "that is how it works at X" could be taken as I mean they should grade on a curve.
Pedantry and intellectual hazing seem endemic in academia -- from the perspective of someone who spent entirely too many years in the sector (all the way through postdoc).
> Pedantry and intellectual hazing seem endemic in academia
I wouldn't call interesting test questions pedantry or intellectual hazing. Giving tons of basic easy questions is pedantry and intellectual hazing to me, I'd rather just have the harder questions.
But on the other hand, yeah, for fragile egos and given the very high stakes associated with one's transcript, not everyone is there to get good and it can absolutely be devastating to never ace tests like you did in earlier years.
Different programs/eductators may have more sympathy for one group than the other, but there's no viciousness involved.
I would understand the third level (possibly even the second level) as including that ability to apply knowledge in nonstandard contexts that the previous poster was talking about - being able to see not just how to perform a calculation, and how to use it in a given context, but also how to create that context when necessary - seeing that the question on how humans move in groups could be modelled as a fluid dynamics problem, for example.
The general wisdom amongst teachers is that most exams should be able to differentiate between these three levels of understanding, so you want questions where the student can just follow the line and do the right thing, but also questions where they're being asked to apply knowledge in different contexts, prove why something is true, or explain why some fact is valuable or relevant in a given context. This isn't about grading on a curve, it's about differentiating between how far different students have come in their understanding. (And the corollary if that differentiation is that if all the students have the same understanding, it's likely that the course wasn't challenging enough, and many of the more advanced students could have been pushed more.)
That said, these regions are more about assessment in general than specific written exams, and the third level will be easier to test in some exam formats than others - a multiple choice exam leaves little opportunity to demonstrate that understanding, whereas an oral exam allows the examiner to explore the subject a bit more deeply. I studied in the UK where a single end-of-unit written exam is mostly the norm, and so these exams had to test all the different levels of knowledge, but if you're gathering results over the course of a year or semester, that might look differently.
Harvard admits quite a few people who would normally not qualify as the cream of the crop. Above average, even good - but the worst Harvard admit is probably on par with the smartest kid at the local community college. They might bring other things to campus! But that doesn't mean you want them handling your legal defense or managing your money.
> and were subsequently successful there.
If 80% of the grades at Harvard are As, this just means you're not in the bottom 20%. I would also generally assume that this 80% is not uniform - some classes or even majors will give out higher percentages of As than others.
Students like this aren't always shy about demanding a formulaic way to get an A. You can try to fight it, grade on a curve, etc, but there's only so much time in a day and in the end it's much easier to come up with a formula that they can follow, and to let them follow it.
No amount of conformity and intelligence can explain 80% A's if there is a curve that mandates otherwise. This grading distribution can only occur if the faculty permit it to, both at an individual level and at a policy level.
A curve requiring an entire class of people who all got 90% of the questions right on an exam to have a particular distribution of As, Bs, Cs, Ds, and Fs would be insane. Curves are supposed to account for "this test is quite difficult and has an expected range of passing scores that doesn't quite fit in the ten point scale" scenarios.
If you can get into Harvard, you're supposed to be an A student.
EDIT: The parent comment makes it sound like admittance to Harvard is prima facia proof that students are already excellent. If that is the case, then why grade students in Harvard? Students at Harvard should be trusted to learn everything they need to learn.
EDIT2: Given that 80% of grades at Harvard are 'A', it seems like Harvard administration agrees that grading is not needed.
If everyone at Harvard exits with a good grasp of Calculus, that's useful information. If only half of people from BU graduate with a good grasp of Calculus, that's also good information. BU and Harvard don't need to have the same proportions of A students for it to be a useful data point.
Incorrect! An A just means you got 90% or better accuracy on tests/quizzes. 60% means you earned enough to move on to the next subject.
Therefore, "What's expected of you" is to get a D or better. Otherwise only an A would be a passing grade.
You're not going to get into a prestigious school with grades like that but community colleges will still take you.
Grades don’t show your ability relative to others, they show your mastery of a set of knowledge and skills. You mastering something doesn’t prevent your neighbor from also mastering it.
https://en.wikipedia.org/wiki/Academic_grading_in_the_United...
In my day, Rank-Based grading was the system, but not just among the students in a specific class -- it was based on historical numbers. The same course was taught for many years so it was easy to get that longitudinal view.
Especially in secondary and high school, a 'C' meant you were average for the students who took the class. The tests had questions of differing levels of difficulty. If you couldn't answer the difficult questions, you clearly didn't "master" the material and therefore couldn't get an 'A'.
People with average ability would be expected to get an average on the test, so a 'C' would still mean you learned the material adequately and could move on to the next class / grade.
Of course, that doesn’t have to be a letter; percentages along with specific corrections would do as well. But once there’s a convention people often use it.
Grading students relative to each other gives different information and probably not precise across cohorts at that.
My AP Physics class had 20 kids in it, 10 of us got 5s and 10 got 4s on the AP Physics exam. Should some of us had our scores lowered for a better bell curve? No; he was just a really great, engaging teacher who pushed us to achieve.
There is no problem if a subset of the total population does very well (or very badly). But if half of all AP takers got 5s and the other half got 4s, then it would be time to reexamine the test.
Similarly, if there was a study group in a class that did very well, that's no problem. But if nearly the entire class gets As, that signals that the class could be more rigorous, cover more material, etc.
I get why they don't do this — they don't want to disadvantage students as they go looking for jobs. But some schools are happy to have a reputation for for tough grading. My college had freshman courses where only 10% could get an A or A- (this amounted to 3 students, since the classes were capped at 30). It's a very selective college, so pretty much all of the kids were "A students" coming in. Among graduate schools and top employers, it's known to be a school that is very academically challenging.
Maybe, maybe not. If you spot check and a statistically significant sample shows kids are indeed learning first year college physics effectively in AP courses, perhaps you do a little happy dance and be glad something's working.
Employers don't seem to be hiring Harvard grads and going "holy shit this person doesn't know anything they're supposed to know" to half of them. Employers seem to be generally happy what an A student from Harvard comes to them with.
> But if nearly the entire class gets As, that signals that the class could be more rigorous, cover more material, etc.
Or that you've got a self-selecting population of high-achievers.
> My college had freshman courses where only 10% could get an A or A- (this amounted to 3 students, since the classes were capped at 30).
That's depressing. "You're an A student, but I'm required to give you a B for organizational reasons" is an absoutely nasty thing to do to a college kid.
That's because they (1) know there's rampant grade inflation and (2) use other indicators to assess candidates. And I've definitely heard people grouse about Ivy League grads, and how they'd rather hire from other places.
> That's depressing. "You're an A student, but I'm required to give you a B for organizational reasons" is an absoutely nasty thing to do to a college kid.
"Organizational reasons"? No. It's because they view grades as a tool to let kids know how they are doing compared with their peers. If you don't want to go to a school like that, there are plenty of grade-inflationary schools out there for you to go to.
A perfect educational system would have every student getting 100% on every test, right? That's the ideal. Unless, of course, that the purpose of the educational system isn't actually to educate people but instead to weed out certain classes of people.
"We want the best" is what every employer will always say but the truth is that they really only need someone who can do the job. We as a society put way too much importance and faith in credentials.
Then perhaps there should have been more available to be learned. If most people are above 90%, then you're holding the best back. (If you're saying that more should be taught, but not be on the test... Why?)
The only way for everyone to reach full understanding at the same time is for everyone to be the same or for some people to have been held back. I view the latter as being far more achievable than the former. I would also say that holding the best back... is probably not what we want the #1 ranked University in the world to be doing.
> A perfect educational system would have every student getting 100% on every test, right? That's the ideal. Unless, of course, that the purpose of the educational system isn't actually to educate people but instead to weed out certain classes of people.
No, a perfect educational system would not have every student getting 100% on every test because not every student is the same and also such a test fundamentally provides no information to the tester.
A perfect system would probably have every student on an individual (or near-individual) learning schedule, and even there I would expect perfectly calibrated tests to average around 50% with adjustments as required for psychological impact or to improve recall. From that, you can identify what elements need to be covered again, and what can be skipped when moving forward because they are already known.
> "We want the best" is what every employer will always say but the truth is that they really only need someone who can do the job. We as a society put way too much importance and faith in credentials.
This is probably correct for your average community college. It might be correct for a average state school.
When you get to 'literally the number one Univeristy on the planet', perhaps employers are also actually looking for the best people on the planet.
There is, and you can go get a masters or PhD if you want that.
> If most people are above 90%, then you're holding the best back.
The Feynmans of the world are perhaps not too concerned with their GPA.
> perhaps employers are also actually looking for the best people on the planet
In that case, they probably shouldn't be looking at grades at all.
This is a ridiculous statement. Let's assume you have a room full of perfect students. You spent the week teaching them about subject X and came up with a test to determine if they got the gist of what you taught.
On Friday you give them that test and every student gets 100%. By your reasoning, because they all got 100% this test was a waste of time and would "provide no information to the tester."
My argument is that the test served its purpose: It demonstrated that everyone understood the subject and you can move on to the next. Could some students have absorbed far more information in that same amount of time? Probably. Is that the teacher's job though? No. Their job is to teach a certain subset of a subject and do their best to make sure every student understands it well enough that they can move on to the next.
The problem with this method is that smarter students get bored out of the minds and end up feeling like they're in prison for their academic career. However, the system itself wasn't made for them. It was designed for the average.
If you feel that grades should be used to separate or weed out students who don't perform as well you're trying to twist "education" into something that it's not: A caste system.
I would say I am trying to return education to its original purpose, instead of the world's worst but also most common form of daycare.
Why? What benefit is there to the world to making students flunk out of a school when that same student would have passed with flying colors elsewhere? "Sorry, you've got a great grasp of engineering that any firm would benefit from, but you'll work at McDonalds now, because Bob got one more question right. Thanks for the $100k and four years of your life!"
The goal of education is to educate, not to flunk a certain percentage of people.
No. The purpose of normal college is to educate.
The purpose of Ivy league school is to separate the exquisite from the mediocre. If you aren't exquisite then why are you going to $100k college? Because your parents can afford it.
There is some low income kid who actually is exquisite who should get that seat in the exquisite class.
You don't deserve to pass just because you're not an idiot. If you are even a little bit of an idiot you should not be able to get a Harvard education. You should flunk out, lose your parents money, and try again in community College.
You are not entitled to pass because you are average and your parents are rich. College is not risk free.
You should be entitled to pass if you can successfully demonstrate your understanding of the subject matter in question. Proposing to curve passing grades down "UNTIL a certain percentage fails" is just... nutty.
> You are not entitled to pass because you are average and your parents are rich.
Yes, which is why grades should be based on "did you learn?", not "do the grades look like a proper bell curve?"
> There is some low income kid who actually is exquisite who should get that seat in the exquisite class.
A low-income kid is the least likely to be able to take the risk of going to a college that's intentionally trying to flunk people out even if they're competent at the subject matter.
This seems to presuppose that the grading system is universal across all universities - i.e. that we would expect the vast majority of Harvard students to get an A, but the vast majority of students at a middling community college to get a C or D.
Is this really the case? I'm not American, but in the British education system, this is broadly considered not to be the case - if anything, the more prestigious institutions historically have had much higher standards for each grade.
Another expectation is that students' performances at university typically cannot be easily predicted from school grades or tests, and that all universities will have a proportion of students who do very well, and a proportion of students who do not do well at all. There is no real expectation that, say, the Oxbridge universities only take students who will definitely get a first, and the other universities mop up. Oxford and Cambridge are definitely more prestigious and therefore harder to get into, but they're by no means some "sure thing" where if you've passed that hurdle, you'll be able to pass everything else.
https://en.wikipedia.org/wiki/Legacy_preferences
Back to US schools, I'd like to see some analysis on the correlation between the proportion of # of GE courses for a program and that program's grade inflation levels. Also, compare public vs private schools' grade inflation. Schools in general have a strong incentive to send their students to internships and post-grad programs at prestigious IB firms, consulting firms, law schools, grad schools, and etc; and I wouldn't be surprised if private schools tend to have more lenient policies in relation to grade inflation to boost that goal.
(edit: grammar)
On a side note the bigger joke are the scholarships for sports - you dont even need to be able to read if they put you on the football team.
You are confusing cause and effect there. By giving everybody high grades, they are reducing the ability to tell Harvard graduates apart, thus increasing the desirability of getting into Harvard and decreasing the need to put in effort after being accepted.
* Minus the legacies and the atheletic admissions I guess.
Well the other 20% needs to come from somewhere!
Grade inflation is real and I think it leads to significant unintended consequences. I do find myself flip flopping however on whether there should be grades at all. If you’ve read zen and the art of motorcycle maintenance, you’ll recall there is a powerful case made to eliminate grades at the college level. I still lean toward having them currently.
Unfortunately, I recall in 1996 4 or 5 students committed suicide at MIT due to the academic rigor (which is why the first year was pass fail).
I’m not advocating for this system just sharing stories told to me.
In my experience, MIT is actually pretty good about wanting to help students work their way through once they've been admitted. But there's definitely a firehose aspect to a lot of it a lot of incoming students aren't prepared for--even leaving aside a lot of the distractions and other elements of suddenly being away at college.
Once we started grading, suddenly the game was to get a good grade. And students have all kinds of ways of getting good grades that don't lead to jobs.
Maybe this is the same as the average across all of the top 20 schools?
It also raises an interesting question: should it be harder to get an A in a class at Harvard than at e.g. a third-tier school? Because the answer to that is not obviously yes.
E.g. if you take Calculus 101 and master it to some level of proficiency that is an A- at some third-tier school, should you also get an A- in Calc 101 at Harvard, or should that merely be a C there? (And if you think it should be a C, then why? Isn't calculus just calculus? Either you master the material or you don't. Why should you be graded differently depending on where you learn it?)
Because if you think grades should reflect some kind of absolute level of proficiency, then it's not surprising at all that 80% of grades at Harvard would be in the 'A' range, considering how many straight-A applicants Harvard accepts in the first place.
An anecdote: my father was valedictorian of his podunk high school but nearly flunked out of his first year of college. And he's a smart guy, it was just a mixture of (1) not having study skills, since he'd never really needed to study before and (2) being a little behind on certain subjects (like math) that hadn't been taught well at his high school compared to the schools his peers attended.
In a situation like that, I'd assume that an A in a class whose teacher is just phoning it in is not equivalent to an A in a class on the same subject whose teacher is serious about both teaching and grading.
IMHO, standardized tests should always be about mastery and scoring should reflect that rather than having a simple number like 1200 or 1600. Someone could get a 600 on their Math SAT but the parts they got wrong were related to simple fractions or time-based calculations. If the only information you have is, "600 out of 800" you don't have that information.
Furthermore they shouldn't even have scores, really. They should be pass/fail based on mastery of required subjects (whatever the test is about). Either you know quadratic equations or you don't. If you got 90% of the quadratic equations correct you know how to deal with quadratic equations... You probably just made a minor mistake somewhere or were running out of time. You still "got the gist of it".
It would make a hell of a lot more sense to me if students had to take dozens of standardized tests demonstrating mastery of all facets of the high school curriculum rather than one great big test at the end of their schooling that measures their ability to remember randomly-sampled fractions of that knowledge.
Wait, if the one big test at the end was randomly sampled well, it would still provide results almost as good as requiring 10 to 30 times the same amount of test time and energy per student. I don't think this cost to benefit ratio makes sense.
Do you really think an intro to calculus class at the local community college and at Stanford should be the same class? Because I guarantee that the standardized class will either bewilder most of the community college students or bore/waste the time of many of the Stanford students.
But one should argue that those good schools should have a more difficult program.
This of course leads to the problem known in Poland (before the 2009 reform): the top schools ha# harder programs. For example if you studied French, all the students allready knew it before joining the university. They learn the advanced topics. Or if you studied maths on the entrance exam they would ask you about topics that are not part of the highschool curriculum. They changed it recently, but Im not convinced that for the better. Now everyone writes the same final exams, each university doesnt run their own exams anymore - at least in theory. Some still do
In 30 years of teaching first- and second-year computer science courses, I always saw a bimodal distribution, with one mode around a C and the second at a B+/A- level. (In first year, it often was trimodal, with the third mode at an F-, students who gave up early in the course and didn't withdraw). I would be astonished if there were not something like that going on in Harvard's first year CS.
Mr Turley quotes William F Buckley on preferring the first 400 names in the Boston phone book to Harvard faculty. I would prefer to listen to a conspiracy theorist explaining why the lepton samurai of Jupiter were about to invade the earth than to listen to Mr Buckley, even if he were still alive.
MIT, Stanford, and CMU do stick out to me as a positive signal but I could care less about anything else, including Harvard.
The case against hiring people from Ivy League schools: https://bigthink.com/business/case-against-hiring-ivy-league...
As much as I dont like grads from top schools. Here they showed some actual smarts. You dont need to work 100 hour weeks in a sweatshop that tries to "balance you between burned out and hungry". Probably that grad has rich parents too, so they can switch jobs to a boring corporate job. They have a safety net not to be exploited with 100hour per week gruel. Come on? Is that some success story? 100 hour weeks is eithet a sweatshop or shitty work organization.
I bet that lady wanted that motived, hard working harvard grad for 15 dollars per hour and also with 50 hours of UNPAID overtime. That's jusr the typical sweatshop approach. Also they should "think big" but not too big like for example "we should pay better in our sweatshop".
I looked who she was - and seems her own partners kicked her out from that company that she was supposedly building by exploiting people.
Harvard average GPA was a 3.0 in 1967, now it's a 3.45 [1].
There are some universities that have tried to actively prevent inflation (e.g. Purdue with a 2.73 average in 1986 to 3.09 in 2012 [2]).
[0] https://www.gradeinflation.com [1] https://www.gradeinflation.com/Harvard.html [2] https://www.gradeinflation.com/Purdue.html
I would have burnt out on my schooling so fast as to make it worthless. How demoralizing it must be when every test is designed so that you won't know all the answers. Why even bother trying? I'm not in school to compete with other students, I'm there to learn.
The most informative grades would be uniformly distributed, it’s an opportunity for the geniuses to shine and give a signal to profs about who to hire for their PhDs.
The only function of giving everyone an A is to prevent shattering fragile egos who have never experienced setbacks before.
Unfortunately real life is full of set backs, so it’s better to learn to deal with it in a safe environment.
A common understanding of a grade is as an indication that the course material has been mastered. Higher grade = more mastery. This is a measure independent of the performance of fellow students.
If you gave a test out and the average was 99 percent, you would be ranking noise. They should make the course harder so even out the distribution of grades.
The most informative distribution (ie maximum entropy) of grades is uniformly distributed. The least informative is when all grades are the same.
There's a bald assertion here that lots of those professors would stop you at; that good grades and capacity for research work are the same (or even necessarily closely correlated).
If a class of 30 physics students all get an A, the value is in comparing their A to a B from another year or university.
There is no way all of them did. Eric Mazur tested this for people taking physics at Harvard, he found that even though every student got an A they didn't actually learn the material. He tested before and after on standard physics understanding, they didn't improve even though they got A in the class.
You can watch this video where he talks about it:
https://youtu.be/tn1DLFnbGOo?t=959
Teaching so that students actually learn the material is extremely hard.
I guess if you have to hire Harvard new grads, want to base your decision on their academic performance, and would prefer more fine-grained rankings of performance, this is not in your interest.
The very top students may also wish they had more chance to distinguish themselves. However the way it seems to work in many schools with grade inflation is that it's easy to get an A, but still pretty hard to get an A+, so there's still some opportunity here.
I'd also add that at Princeton, attempts to combat grade inflation were very unsuccessful and resulted in strange unintended consequences, like students carefully strategizing which sections of courses to take to avoid those believed to have heavy competition.
Certainly, if you're going to major in some quantitative field, you can only avoid so many of the courses in that major. But, if you're really interested in optimizing for GPA, you can often still major in some variant of the major that provides more flexibility and--where you have electives--take classes that are easy (for you).
If you're already in the top 10% of all students in the USA, you can still be average in comparison to your peers.
A 'C' grade at Harvard should still be pretty good in the job space because: 1) you were good enough to be admitted to Harvard, and 2) you are representative of an average student at Harvard.
I've never heard that before, and it certainly hasn't been in any class I've ever been in. And Wikipedia says nothing of the sort:
https://en.wikipedia.org/wiki/Academic_grading_in_the_United...
To the contrary, C is generally the lowest grade that is considered to be a satisfactory passing grade.
Says who? I find mastery-based grading much more logical.
Take the Finnish or Estonian school system, it's very rarely curved like in the States. If you master the material you get a good grade. PISA results seem to indicate that works well.
AP score data is especially indicative of this. In AP classes, most high school students earn As from their teachers (I estimate around ~60%). Yet, when you look at the AP scores, few get 5s on exams (the equivalent of an A). For example, 5-10% get 5's on AP English and Science-related exams. Scores are a bit higher on Social Sciences (10-15% 5s) and math/CS (~25%). But only ~50% of students even get above a 3 on the exam (the equivalent of a C). So there are people who essentially get a D by the standard (a 2) that are getting an A in their high school class.
I could go on with countless examples of how students aren't nearly as capable as their grades would indicate. At McKinsey, I interviewed over 100 people from top schools with high GPAs. Many couldn't solve simple math problems when given the problem in the context of a real-world case.
At out top institutions with our top students, we should be pushing them extremely hard and measuring them against a higher bar. I find it disappointing that they are being measured against the same (or even a lower bar) than students at other institutions. It's so bad that it seems clear that a Harvard Education isn't any better than a Penn State, Nebraska, or even UC Riverside education. The only difference is having the Harvard brand and network. It's an embarrassment. And it's not just a problem at Harvard. It's a problem at every top institution.
Grade-motivated students don't put in the extra work when little effort still earns them As. One example stands out. I run an edtech company, and in our early days, we ran an experiment at doesmyessaysuck.com. You could submit your essay, get a score, and 2 pieces of feedback on how to improve the content and structure of your writing. We did about a thousand essays before abandoning it. I always think about one student. He submitted an essay we generously gave a C (it was really bad). It was incoherent, poorly structure, didn't answer the prompt, and lacked sound logical reasoning. He responded we were wrong because he had gotten a 93% on it.
Also – https://www.gradeinflation.com/ is a good read. It's out-of-date data. But it proves the point even further than this article.
Note: There is some research out there that claims grade inflation is false and that more students actually have learned more than in previous years (i.e., the academic bar hasn't changed, but more students are above it). However, this research is unconvincing (and ripe with errors). There's essentially been zero improvement (and even a decline) in basic math and literacy skills over the past couple of decades (as measured by standardized tests). Yet grades are much higher. And high school and college graduation rates are up by about 10 percentage points since the early 2000s. As it turns out, when your only measure of success is graduation rate, you end up with more graduates – even if those graduates don't have any skills.
There's absolutely no problem with this.
Your grade in your AP class is a measurement of your achievement at the high school level. An A indicates you did what's expected of you as a junior/senior in high school.
The AP exam is measuring whether you achieve at a first year college level enough to skip that first year in the subject.
More people should get As in AP Physics than get 5s. If they measured the same thing you wouldn't need the AP exam.
They are high school classes at an honors level that may generate college credit. A two on the exam is deemed “possibly qualified” to skip a year of college, for example, but colleges won’t take it. Many won’t take the three score’s “qualified” either.
For purposes of your GPA and high school physics, that’s not a D.
The AP physics exam gives you a 5 if you miss like half the questions, too. That doesn’t fit a ten point GPA scale very well.
Q: What do you call a doctor who graduated at the bottom of their class? A: "Doctor"
That said, I agree with other commenters that you set this situation up with the rigorous acceptance criteria. You're testing the people who got into one of the best schools so presumably if your selection criteria works they should all excel at the courses.
From my experience, college students worry enough about grades without forcing the curve to be harsher than it currently is.
You can give out super hard exams and make them sweat/cry/worry, but you'll be given tons of shit if you actually fail them, especially if their parents are large donors. Only exceptions here are for students that failed so badly they're no way they can pass, even if you apply a crazy curve on top.
now many (most?) of those students are extremely talented and deserve the grades they get, but it's still a bullshit system that encourages failing upwards.
I did my undergrad outside the US but in a university that follows American system. My department culture was that you don't expect curves and grades are based on what you get. So if we are all getting F then that is F for all people. No one passes a course just for the sake of passing. While these extreme cases never happened (at least when I was there). if it happens then this tells about quality of the course itself not only students effort and performance.
I took a general relativity class during my undergrad. I like this field and topic. And while I am currently doing particle physics but I always liked GR. I was doing well regarding the topics but never got good grades. I struggled with its exams in particular with this particular professor. I got C although I managed not only to have a good grasp of the subject but went so far as to work on specific problem and expanded it into a paper, participated in a conference and got it published later into a peer-reviewed Journal.
While when I moved for a graduate school is the US. Things is very different. Many of the undergraduate I teach were always expecting high grades anyway. curving is very much the norm and the negotiations is about how much to curve. The normal distribution of the grades is a religious belief to some people in the universities administration. And I understand it is hard to assist professors teaching and some use this as an evaluation metric but it I feel like it doesn't addressa core thing. Some people will pass even with high grades and they did not get the core of the subject. This is not about they deserve more than me or not. There is a fundamental question of are we grading on effort to learn or actually what people learned. The grading criteria will differ much based on our answer to this.
In your profession effort to learn new thing is the most important thing. But with the degrees that assumes that you are ready if you have it is that the expectation or your effort and ability to learn. But how can wr measure that specially on scale.
During my graduate courses I took a QFT class taught by condensed matter physicist and was focused on his area of research. I wasn't interested and it was very tough. I don't feel I understood most of the things and I struggled to the point I did not submit the final exam (take home). And to my surprise I got an A because I was the highest grade in my class (we were few anyway). I learned a lot of QFT later on my own but I am giving this example on two completely different personal experiences that hopefully deliver the idea.
It is hard to design a grading system on a scale that will try to measure different things. and the universities marketing for degrees is doing much harm ( for financial reasons). And also graduate schools and companies don't usually want or sometimes doesn't have resources to look into details about applicants institutions even if they have information about what their grading works. Hell I don't even think it is possible to have one unified system per university (Is it even good thing to do?).