Hacker News Reader

Comment by kanobo | Hacker News Reader

Maybe the industry could popularize a 'normalized-grade' in applications where it simply asks for the student's grade divided by the average GPA at their school? Is that a bad idea?

View on HN

My objection to this is that it assumes cohorts are equivalent.

Also, you have now incentivized zero sum competition within the cohort or finding a weaker cohort to beat.

I think the world is more complex than that, people would still rather go to a good competitive school rather than a school of weak peers. It fosters higher scores and standards in global activities such as standardized tests and national awards, etc. If this wasn't true, bad schools (pre-college) would be much more popular since it is much easier to get a higher GPA.

The idea that the perceived value of something is equal to its actual value is far from reality.

I have met plenty of people that went to top 20 schools that just gave up once they got there yet, still managed to graduate and go to graduate school. I have met plenty of people from my undergrad (definitely not a top 20 school) who were absolutely relentless in their studies and managed to have far more lucrative careers than those slackers who went to top 20 schools.

I mean, I have nearly 3 degrees in EE, CS and Physics.

Physics is usually one of the hardest majors at any university and my homework assignments/tests weren't much different than MIT's homework assignments/tests; sometimes they were a lot easier and sometimes they were on par with their graduate classes. For my lectures where MIT had their OCW version online, they were near identical; the lectures where there wasn't a corresponding OCW lecture, the quality was around the same. Moreover, CS and EE weren't much easier than physics so I can't imagine top 20 universities had much higher quality there either.

(My university also deflated grades to the point where the average GPA was a 2.7 which made it that much more cut-throat (since a 3.0 was the cutoff for many engineering jobs))

(Also, from what I recall, UCSB was much harder to get into than my school yet, they were blacklisted at many engineering companies for having the earned reputation of being a party school; while, at my school, we had those companies consistently come to our career fairs)

If idiots manage to graduate top 20 schools and yet idiots of similar caliber can't graduate from schools that are far below the top 20, their grade inflation is completely unwarranted.

If you're interested in this sort of thing, check out the R-score used by CEGEPs in Quebec. It takes into account how well you did relative to the average, what the standard deviation of that average was (the lower the standard deviation, the more significant your deviation from the average is), and then how strong the group of student that made up that average was (group strength is based on students results from standardized exams taken in grades 10 & 11). https://en.m.wikipedia.org/wiki/R_score

So Chem majors have their grades adjusted according, in part, to the the grades of Ed majors? I don't believe that is an improvement. I rather see that likely to result in grades shooting up as majors seek to beat the curve.

And if we adjust Chem majors according to the grades of Chem majors then small majors, with a high standard deviation, would have greatly different adjustments from year to year (and at my school there are many majors where the numbers of grads in a year is less than 10).

I think you're right in that it wouldn't be as meaningful for a school with departments with only 10 grads, but I think it's unlikely that a department's chair and faculty would care or even consider to mandate their grading to be curved in an effort to raise their student's GPA to compete with other departments. A difficult major is expected to have a lower average GPA than an easy major in any school, that's why latin honor thresholds are different for each department and different majors don't get considered for the same jobs purely on GPA.

No offense, but I can see that you don't work in a college or university, or for that matter have experience with Gresham's Law.

No offense taken, but I'm honestly an Associate Professor at a major research university. But no worries, I respect your viewpoint.

Ah, I see. Give it a few years, then.

If everyone has A, it still leads to A.

If you have a 4.0 but your school's average is a 5.0, your normalized gpa would be a 0.8 vs a 1.0. I didn't mean to imply that the divisions would use graded letters.

If everyone has 5.0, the average is 5.0. With even more pressure, since even smallest downgrade to someone has huge impact.

I did not meant to imply you would average letters, they are just the usual way to express the grade.

I think we have different priors because I definitely didn't have the best grades compared to my peers as a student. And as a teacher at a University my students have a huge range in their scores.

If a C is what the average should be, then the A's would all get normalized to a C.

The algorithm is supposed to be "simply asks for the student's grade divided by the average GPA at their school".

Yours would put students at schools with other good students at disadvantage against those who go to less competitive schools.

The assumption that students in different schools should have the same average is likely wrong.

If you divide by the average score at the school, it assumes professors grade on an absolute scale that matches across universities, that a 4.0 is the same anywhere. Since this is not the case, then there already is built-in bias to the 4.0 at a certain school. At least if it is normalized to a C average, you'd know how a student stands compared to his classmates. You can then scale by the approximate academic rigor of the institution. If you don't remove the DC offset to the grade, then you would need to know both the individual's GPA along with the average GPA anyway to understand the number. This normalizing to a C average saves a step.

Reply on news.ycombinator.com