Adjusting GPAs: A Statistician's Effort To Tackle Grade Inflation

An anonymous reader writes "A recent analysis of 200 colleges and universities published in the Teachers College Record found 43 percent of all letter grades awarded in 2008 were A's, compared to 16 percent in 1960. And Harvard's student paper recently reported the median grade awarded to undergraduates at the elite school is now an A-. A statistician at Duke tried to make a difference and stirred up a hornet's nest in the process."

Comments

Use Class Rank

[the eric conspiracy]

Ignore GPA.

Re:Use Class Rank

[rritterson]

The essence of class rank is to compare the student to his/her peers instead of against a fixed measure whose bar can be raised/lowered.

Class rank is problematic though, for a couple of reasons:
-It doesn't make sense to compare GPAs across majors. The article points out that natural science professors already grade more stringently. Class rank across the entire university would only ensure natural science students looked poorly. (And vice versa for humanities students)
-If your GPA is going to be directly compared against others as a measure of your talent, you have an extra incentive to find a way to take the classes offered by the professors who grade most easily, boosting your GPA and thus your class rank.

We actually have a time-tested way of comparing students' performance to each other: grading on a curve. When I was in college (early 2000s, major American public university), all science and math courses were graded on curves, with 10-15% of the class getting As. Most professors had a minimum score that would guarantee a passing grade so that there wasn't a necessity to fail anyone, typically set to some percentage of the median Some students complained that they were doing well and learning the material, but are only getting Bs because of superstars in the course. To that, I say tough, because in the real world, no one is going to hire you to do anything just because you are good enough if another candidate is around who will do a better job than you will.

Fortunately, my university's grading policies were well known enough by employers in my field, so that the relatively lower GPA were taken into account when recruiting. The best students had A's in about 2/3s of courses. Hardly anyone had a 4.0 in even a single semester, just because it's extraordinarily difficult to be in the top 15% in every subject and have any kind of regular life.

Re:Use Class Rank

[fredprado]

Ignore both GPA AND Class Rank. Let the graduation schools apply entry tests. Problem solved.

Re:Use Class Rank

[Nemesisghost]

Class rank is problematic though, for a couple of reasons: -It doesn't make sense to compare GPAs across majors. The article points out that natural science professors already grade more stringently. Class rank across the entire university would only ensure natural science students looked poorly. (And vice versa for humanities students)

The easy fix for ranking is compare each to those within the program they belong. For example, only CS students would be ranked against other CS students. For that ranking only compare courses required by that degree, so those taking the humanities major vs those who take a math minor will be ranked the same. Double major? Double ranking, one for each major.

As for actual grading each course, I like the idea that one of my professors used. He graded each assignment & test with no curve, and only gave us raw scores. Then to determine the course grade he looked at the grouping of grades. There was always a break between the A's, B's, C's, etc, and that's how he determined who got what grade. I don't think he actually failed anyone(or gave less than a C) unless they simply were not trying(ie skipping class, not doing homework, etc). He was the Dean for the College of Science & this was a physics course, so nobody ever complained about the grade they got. I think most in physics department ended up doing something similar to this and it worked out really well.

Re:Use Class Rank

[SJHillman]

Not to mention class rank is heavily influenced by the size of the class and breaks down in small courses/schools/etc. If you go by a single course, you could be ranked #6 and still be at the bottom of your class. Alternatively, you could be in the bottom 25%, but have a 99% average. The real problem, like credit scores, is trying to reduce a complex issue with many variables that are completely out of your control and cram it into a single number that supposedly describes you.

Re:Use Class Rank

[wanax]

Grading on a curve only works for large, introductory courses. The problem is two fold 1) smaller classes cannot be assumed to have a normal distribution and 2) Once you get past intro classes in any subject, there is a strong selection bias so that people in upper level classes all tend to be high level performers in that subject (which also means you can't assume a normal distribution).

The big problem with grades is that they conflate course difficulty and student performance. If you want grades to be a proxy for performance, you have to weight them somehow or other by class difficulty. The problem is nobody can agree on how to rank class difficulty due to academic politics, since nobody wants to be the department that gets the short-end of the stick with class difficulty rankings. In my personal experience, being one of the few people who have taken multiple graduate level classes in 3 disciplines (History, Mathematics and Neuroscience) at that level no field is particularly easier or harder than another, it's just that the type of work one does is very different.

The other issue that I rarely see addressed in all of the 'grade inflation' concern (and which class rank also ignores) is that maybe today's college students are actually working a lot harder than those in 1960 (perhaps due to debt, the weak economy, lack of security from getting a degree etc), and have actually earned a big chunk of the upward grade adjustment. That's certainly been my experience when compared to my own cohort, and that of quite a few professors that I talk to as well.

Re:Use Class Rank

[khasim]

We actually have a time-tested way of comparing students' performance to each other: grading on a curve.

That only works when MULTIPLE RANDOM items are compared. Such as rolling 3d6.

Since answering questions on a test should NOT be random there should not be any reason to attempt to force the scores into a curve.

When I was in college (early 2000s, major American public university), all science and math courses were graded on curves, with 10-15% of the class getting As.

I started college in 1983. The grades were based upon how many questions you answered correctly. It did not matter what other students answered. Why would it?

Some students complained that they were doing well and learning the material, but are only getting Bs because of superstars in the course. To that, I say tough, because in the real world, no one is going to hire you to do anything just because you are good enough if another candidate is around who will do a better job than you will.

By that logic, a "B" student in one class could be an "A" student in the same class with the same professor on the same material with the same answers ... but in a different semester/quarter.

Which means that the smart students will learn to "game" the system.

Re:Use Class Rank

[UnknownSoldier]

> Some students complained that they were doing well and learning the material, but are only getting Bs because of superstars in the course. To that, I say tough, because in the real world,

How is the hell is that fair to steal a legitimate grade from people who earned it simply because you want the grade to be Relative to others?!?!
The WHOLE point of a grading system is to have an Absolute measurement system!

That is you, 50% means you only know 50% of the material. A 100% means you know 100% of the subject. Not, gee, you know 95% of the material but since 10% of the people know more then you we will lie and say you only know 90%.

Fuck You and your Grade Theft aka "grading on a curve".

In the real world most people don't give a shit if you got a C or an A in a course, only that you put in the required effort, you are certified as knowing the material, and the university / college has proof of your efforts.

--
Piracy === Disrespect.
Piracy =/= Theft

Re:Use Class Rank

[Anonymous Coward]

So your method is going to rank math majors against linguists based, ultimately, on which group is better at the History of Art?

Re:Use Class Rank

[unrtst]

The big problem with grades is that they conflate course difficulty and student performance.

IMO, there's also another glaring flaw in Johnson's premise that students gave better student evaluations of teachers who graded more leniently. There is a HUGE assumption there that the various teachers running the same classes were all equal in their quality of teaching. Why is it so difficult to believe that some teacher was able to reach and educate more of his students than someone else?

Statistically, I understand there should be some sort of even distribution, but the sample size (in number of teachers per course) is not large enough to be of statistical value.

Johnson said, “As you might expect, the effect of either expected course grade or received course grade is very powerful in student evaluations of teaching. If a student was getting a C in a course, he or she was very unlikely to rate the instructor highly. If they were getting an A in the course, they’re more likely to rate the instructor highly. I think this provides quantitative evidence for something most instructors know: If they grade easier, they will tend to get better course evaluations.”

One year, I had an art history teacher whose class was at 7:30am, in a dark lecture room, with a dim projector on the whole time, and spoke through an ancient 3" amplified speaker with an voice that was already monotone and droned on and on and on. I got a D-. The next year, I did more research on the available teachers, and found the one that engaged the most and who had more people getting higher grades. He was fun, taught in a well lit class in mid-day, involved us in projects to learn (ex.create an interactive presentation of some artist with a group of other students for homework, as opposed to filling in the blanks on a test in a dim room with projected images), and I got an A+, go figure.

I'll admit his tests were slightly easier (fewer exact date type questions (what year was this created, versus during what time period), multiple choice on name questions, rather than having to fill it in spelled perfectly, etc), but I also learned a LOT more, and neither graded on a curve.

I also take some offense to applying various curves or rankings etc to students. Given a class of 30 people, it's almost guaranteed that you'll have some years where half the class are "A" students, and some years where there's hardly a one, and that's assuming that the teaching and material are equal.

Overall, I'd agree that there is grade inflation. Jacking with the grades isn't going to fix all the underlying problems, and it will create other problems. He notes that one of the most likely reasons are student evaluations - so untie those from teacher review (instead, to review a teacher, do so as one should for reviewing any employee... go watch them while they're actually working, and only use the reported figures to identify people that should be reviewed first or may need help).

Re:Use Class Rank

[ComputerGeek01]

How is grading on a curve better than a strict percentile rank? Is there any benefit to the complexity it adds?

Answer 1: It isn't. Answer 2: There is no benefit anywhere regardless of what you compare it against.

Teachers who grade on a curve don't understand what a GPA is meant to represent. They are taking something that actually a representation of a students performance in a course (since the raw score is an average of their scores in assignments, tests and projects for the class) and they are trying to hammer it into some half-assed solution to compare students against one and other. They are saying "It doesn't matter how well you have demonstrated your understanding of the course material, or how well you've done in regards to the individual assessments that I myself assigned and evaluated. Instead of getting out from behind my desk and developing a system that reflects what I am trying to show here, I'm going to deprecate your grade until it suggests that you have a less then basic understanding of the material that was taught. Because that's just easier for me to do."

Have any of these Gen X retards even considered the students who, despite knowing the material, will have to sign up for, pay for and hope to find an empty seat in a class all over again just because they didn't try hard enough to impress their narcissistic teacher? Probably not because that wouldn't be helpful in stroking their ego and making them feel more important in the world. What actually needs to happen is for teachers from all schools and disciplines need to sit down, STFU and realize that outside of the classroom they have no authority, nobody gives a damn about their opinion and that even those glowing recommendations that they wrote for their favorite students mean slightly less then whether or not the applicants socks match this morning. Their job is to teach and evaluate their students understanding of the material, that's it. It is not to try and decide if one student is better then another or if Little Bobby Tables needs to apply himself more. You want a way to compare potential employee's? It's called a fucking portfolio.

Re:Use Class Rank

[ShanghaiBill]

Good point

No, not so good. Class rank favors those that took easy classes, and those that went to bad schools with lots of dumb classmates.

Here is a better solution: Test for the actual skill you need. If you are hiring someone to write C code, then give them a programming task that they should be able to finish in an hour. If you are hiring someone to dig ditches, hand them a shovel, and see how much dirt they can toss out of a hole in 20 minutes.

Number of times anyone has ever asked for my high school GPA: 1 (the college I attended)
Number of times anyone has ever asked for my college GPA: 0

Re:Use Class Rank

[bondsbw]

I have to agree with the AC. This method would give higher scores to "jack of all trades, master of none" than to those who are especially gifted at certain areas but uninterested in others. That's not to say that it should necessarily be the opposite either; both types of person are important in the world.

But more importantly, why do we feel we must compare math majors with French majors? They have little to do with each other. Even if a few candidates were trying to get a job where both skill sets were important, the employer would look at the corpus of work in both subject areas, not a single number trying to describe both concepts.

Re:Use Class Rank

[EvanED]

IMO, there's also another glaring flaw in Johnson's premise that students gave better student evaluations of teachers who graded more leniently. There is a HUGE assumption there that the various teachers running the same classes were all equal in their quality of teaching. Why is it so difficult to believe that some teacher was able to reach and educate more of his students than someone else?

I can't speak to that, but I will share another "student evaluations somewhat incentivize the wrong things" bit I've seen discussed in a couple papers. One of the big buzzwords in teaching is "active learning", i.e. actually having your students do some work to figure things out rather than just lecture at them. There's good evidence that heavily incorporating active learning does tremendous good for student learning; if you look at students' long-term recall (e.g. in the next course), even relatively mediocre teachers (still measured by their students' long-term learning) who do a lot of active learning probably do about as well or better than the absolute best lecturers.

What happens during student evaluations? Students complain "the teacher isn't teaching" and "why do I have to learn everything myself", and the teacher takes a hit on the evaluations.

Re:Use Class Rank

[PRMan]

I taught Pascal programming at a university as an adjunct professor. I had 7-8 students in my class at a time. There was nobody to fail because they each semester they mastered the material well enough for at least 5-6 students to get A's in the class. They wrote programs that worked and accomplished the stated goals and they turned in their homework and did well on tests.

So I'm supposed to arbitrarily fail somebody that succeeded just to match some inane curve? No, I taught them a difficult subject and helped them until they succeeded. Now, every student I have has been involved in the computer industry (despite this being a small Christian university known most for ministry, teaching and psychology/counseling) either in software training, QA, high school programming teachers, technical writing, etc. Because I taught them something hard, helped them succeed and gave them the A they deserved when they worked their tails off for a 100-level class.

Re:Use Class Rank

[PRMan]

All the long-time professors that I talk to, including one I had who is still teaching 20-some-odd years later and my uncle who recently retired say that students are getting dumber every year. There's no denying that their selective memory only remembers the best students, but they report being completely unable to have the sorts of class discussions that they used to have only 10-20 years ago. And they can tell students exactly what will be on the test including the exact question in statement form and still half the students will fail the test, whereas 10-20 years ago 80% or more would pass easily.

And let's face it. How many of us could even hope to pass this exam today: http://graphics8.nytimes.com/p...

(Although I have since learned that it was a week-long open-book test.)

Re:Use Class Rank

[supercrisp]

In my experience (just over 20 years teaching English courses), students in the first two years of college are horrible now compared to 10-15 years ago. At upper levels, students are indeed working much harder. But part of the harder work is a lot of flailing around because most bright students have never had to study, organize, research, or any of the basic scholarly skills. School has just been so easy. ---- I'm reluctant to address grade inflation on slashdot because so much of the discussion on teaching here is from only the student perspective, and typically from disaffected students who see education as some sort of market exchange. It's got a much older set of models, and that complicates the hell out of things. For reasons good and bad, faculty tend hang onto some Medieval ideas like mentoring, patronage, whipping people into shape, and separating wheat from chaff. As I said, good and bad reasons. ---- But major influences on grades just don't come up in these discussions, so I'll offer two: retention and rehiring. Administrators and evaluating bodies continually yell "retention." What can you do if all your students suck because they're getting shit for a high school education? Dumb down the classes and pass them. Or don't, and your department suffers. Or you do. That brings me to rehiring. Many classes, right on up to the senior level are now taught by "contingent" faculty--the majority of faculty now are contingent. Nontenured. Rehired year by year. If you're contingent, you'd better listen when someone howls retention. And you'd better make damn sure that little Pauly Privileged doesn't go running to your chair bawling because he got a C for his paper copied from Wikipedia. Better give that brat a B so that you can keep paying your student loans. Presto! Grade inflation. ---- There are other reasons. And I know everyone here is super brilliant and earned those A grades.

Teaching is a social "science"

[HellCatF6]

Teaching as a discipline is one of many social sciences,
but since it's not a true science, there is no pressure to
create quantitative measures for any of their components.
No rigor, no quant, and you leave it up to individual motivations
as the driving forces.
Result, as the article states, easier classes mean higher grades.
Higher grades means better teacher evaluations.
Better evaluations means easier job and more money.
Result - grade inflation.
It seems obvious now, so we shouldn't be surprised.
The real question should be this: when can we expect the bubble to burst?

The whole system needs to change

[CastrTroy]

I would rather have a large number of people get A's, and just have people realize that there are limits to what can and should be tested in school. Either the test is made so hard that only a small percentage of the students are able to answer all the questions, thereby making the median grade a C, or we must accept that it's possible that a high percentage of the class will learn everything they were supposed to learn from the class, and therefore receive an A. The purpose of school isn't to differentiate between who are the elite and who are the median, but whether to certify that you learned whatever it was they were supposed to be learning. I know people who have had teachers tell them they won't give out any A's, which ends up being because it's an easy course, and they don't want all the marks to end up being A, because it looks bad, and would rather just give the entire class low marks.

Re:The whole system needs to change

[Max Threshold]

"The purpose of school isn't to differentiate between who are the elite and who are the median . . ."

Yes, it is.

Re:The whole system needs to change

[Rich0]

The purpose of school isn't to differentiate between who are the elite and who are the median, but whether to certify that you learned whatever it was they were supposed to be learning.

The only reason employers look at grades is to judge who is elite and who are the median.

When you get 400 applicants for a job, chances are that 350 of them can do the job. The employer wants the best person for the job, in the hopes that they will do the job better than whoever their competitor hires and give them an advantage.

Saying that colleges shouldn't give out grades is like saying that amazon shouldn't post prices on their website. Instead you should tell them how much you have in their bank account and what you want, and they'll tell you whether you can afford it. If you can, its yours and they just deduct what they feel it is worth. After all, the exact price doesn't matter, only whether you can afford it.

Re:The whole system needs to change

[scamper_22]

I'm also often amazed how people miss this rather obvious point. So much of education IS to differentiate students. I wouldn't say it's the whole of it, but it's a very big part of how our society operates.

Who gets into med school?
Who gets into law school?
How do you justify some jobs getting paid more than others in areas that are not ruled by the free market (governement jobs, professions...)
Who gets some great grad school spot
Who gets a professional job after graduation?
Who gets the high end law articling position? ...

All these things are very much based on education and what grades you get.

Take that away, and 90% of the population would end up being a doctor/lawyer and no doubt we'd introduce some silliness to stop that from happening.

Too many people like it inflated

[edremy]

Students love grade inflation because they love getting A's

Faculty love grade inflation because they spend less time dealing with pissed off students and helicopter parents

Administration likes grade inflation because it means fewer people drop out, which is good for the bottom line. More degrees with honors sounds great too.

All we need to do is fix students, faculty and the administration and we can solve this problem right away.....

Re:Too many people like it inflated

[i.r.id10t]

I recall an article I read 10-12 years ago about grade inflation, and how it really started in the 60s as a way for the "liberal" professors to help keep kids out of the draft for the Viet Nam War. High GPA (3.0 or higher IIRC) let the students keep their draft deferrments, so a lot of instructors were happy to fudge the numbers upwards just a tad.

Re:Seeing past how you count

[SJHillman]

I graduated high school in 2006 and got my Bachelors in 2009. College admission was, and still is, the only thing that even gave half a flying fuck about my High School GPA. Grad school admissions have been the only thing to give a half-flying fuck about my undergrad GPA (and even then, as long as it meets their minimum requirement, they don't much care). Employers have mostly only cared about whether or not I did graduate. I've seen a number of (accredited) graduate schools that only assign pass/fail to courses and don't do GPAs at all. For the most part, your GPA is like your SAT score... it's relevant for a very short time frame afterwards and for a very small number of situations (mostly admissions and scholarships) and nobody gives a flying fuck after that.

Re:Of course students want the "easy A"

[plover]

When my son entered high school, the principal gave parents a talk on Advanced Placement (AP) courses and college admissions. Someone asked "Isn't it better for my son to take an easy class and get an A than to take a hard AP class and get a B?" The principal replied: "it's better for him to take a hard class and get an A, because those are the people he will be competing against."

Good advice.

What philosophy of Education are you using?

[ralatalo]

There is a basic point missing in that expected grade distribution is very much dependent upon if you are trying to teach a subject to mastery or teach a subject the students limits of understanding. Ie. what is your philosophy of education?

If you are teaching a class covering a subject which can be mastered, then there is no reason everyone should not master the material and get an 100% baring lazyness.

An example would be written test for a drivers license, is there really any reason everyone who takes it should not get 100%?

If you are teaching to a scale, then you don't really care how much absolute material is transferred and your tests are designed to not to measure the material taught in the class as much as then general subject matter which the class covers, and they are designed to test the level of understanding of the subject as a whole with an emphasis on trying to prevent anyone from mastering the test.

Most of your Engineering classes.