Grade Inflation in the US

[ImTestingSleeping]

Hi xkcd goers!

I searched the forum briefly and couldn't find a discussion focused on this (which surprised me).

I know that my little undergraduate school's average student GPA is a B+. When I heard this I wasn't surprised, but the figure is staggering nonetheless. It isn't just that professors are giving out A's and B's like their favorite song is Mammia Mia, but also that they're teaching courses and creating curricula which caters to an above average average (I'm assuming due to administrative and social pressures).

What the hell is going on here and why aren't more people talking about this? I'm sure I'm not attending the only college where this is happening. In the end, is this about institutions making their "customers" happy by handing out the grades they want, but not what they deserve?

[The EGE]

In the end, is this about institutions making their "customers" happy by handing out the grades they want, but not what they deserve?

Yes.

[ImTestingSleeping]

In the end, is this about institutions making their "customers" happy by handing out the grades they want, but not what they deserve?

Yes.

That's the case for private institutions mainly I suspect and I see how that could be a very uphill battle to change, but what about state institutions?

[achan1058]

I would think it will be like money inflation. Things will get normalized down. As long as they are still keeping the relative order of the students in place, it shouldn't pose too much of a problem. If a certain school gives too high a grade to students relative to other schools, the school's "value" will go down. While I would wish school just give grades with \mu=70\%, \sigma=10\%, I doubt it will ever happen.

[doogly]

I am more than happy to give everyone in my classes an A. I tell them what I want them to be able to do. I say, "problems like these, but a little easier with a lot less time to solve them in." You take this seriously and practice and you should do well. None of my exams have any sneaky trickses or surprises. It is always very transparent what my goals are for the course, and I make it clear what should be focused on. With such an approach, it's a wonder they don't all get A's!

[Magnanimous]

My Linear teacher this quarter said "If you really need a 4.0 to get into [graduate program/transfer school/etc], just tell me and I'll probably give it to you. We all know grades are bullshit anyway."

In the end, is this about institutions making their "customers" happy by handing out the grades they want, but not what they deserve?

Yes.

That's the case for private institutions mainly I suspect and I see how that could be a very uphill battle to change, but what about state institutions?

Yes. There aren't many standards/regulations about how to "grade" most classes, unfortunately.

[scarecrovv]

I've always found the common Percentage->Letter Grade correspondence (90=A, 80=B, 70=C, 60=D, less=F) to be entirely silly. It only makes sense for a test composed of true or false questions. In that case, I can see where the correspondence comes from (guessing gives you an F, increasing accuracy from there gets you a correspondingly better grade), but it's silly nonetheless, since well over 90% of points are not assigned by true or false questions. Grading on a curve is silly too. If you did a fantastic job teaching, and all the students really understand everything extremely well, everybody should get an A. The way it should really work (and occasionally does, with good professors) is this:

Write lots of questions of varying difficulty, and assign point values according to their complexity. Decide which questions even someone who barely passed your class should be able to answer. The total number of points of these questions are the D cut-off. The points from the questions someone who passed comfortably (but didn't really do very well) should be able to answer are the C cut-off. The B cut-off is for students who did a good job. The A cut-off includes all of the questions that require deep understanding.

It requires that thought be put into it, but the end result is far more fair and less arbitrary. Obviously announce what the cut-offs are, and the point values for each question.

[achan1058]

I've always found the common Percentage->Letter Grade correspondence (90=A, 80=B, 70=C, 60=D, less=F) to be entirely silly. It only makes sense for a test composed of true or false questions. In that case, I can see where the correspondence comes from (guessing gives you an F, increasing accuracy from there gets you a correspondingly better grade), but it's silly nonetheless, since well over 90% of points are not assigned by true or false questions. Grading on a curve is silly too. If you did a fantastic job teaching, and all the students really understand everything extremely well, everybody should get an A. The way it should really work (and occasionally does, with good professors) is this:

Write lots of questions of varying difficulty, and assign point values according to their complexity. Decide which questions even someone who barely passed your class should be able to answer. The total number of points of these questions are the D cut-off. The points from the questions someone who passed comfortably (but didn't really do very well) should be able to answer are the C cut-off. The B cut-off is for students who did a good job. The A cut-off includes all of the questions that require deep understanding.

It requires that thought be put into it, but the end result is far more fair and less arbitrary. Obviously announce what the cut-offs are, and the point values for each question.

The issue is, profs usually have no idea how difficult their test is. This is particularly true for physics/engineering/math profs. So much so, in fact, that we (TA/profs) test drive the midterms by assuming we have anything from 1/2-1/10 of the time that the students have. Unless your questions are designed to be simply drill questions, you are going to have a problem. Prof can be like "Oh, this is cute, and probably a little challenging." and the result will be murder. Trust me on this, I pulled a question on the book and thought "They only need to copy the proof I did in class and modify a couple of lines.", and approximately 30% of the students got an idea of what to do on the HW. (with < 10% actually getting doing a good job) Granted, I don't have much teaching experience, but I heard that older profs tends to do worse in this area.

[++$_]

The problem is that students have incentives to lobby for higher grades, while professors do not have an incentive to resist these demands. If a professor decides to lower the average grade, this will lead to poor evaluations, which at least theoretically could hurt the professor's career, and few professors are willing to take the risk.

[ImTestingSleeping]

If a professor decides to lower the average grade

I think you have to be careful with a statement like, "If a professor decides to lower the average grade". The actual grade inflation is a result of the true problem in my opinion, which is that students are allowed to maintain a GPA which doesn't accurately reflect their understanding of the material. It isn't a matter of lowering the grade average for the sake of it. I'd be fine seeing a campus with a 3.0+ average GPA if that was a good indication of where they were academically. That just isn't what is happening.

However, I think you were implying that the professor make the class more rigorous which would most likely result in a lowered average GPA.

The problem is that students have incentives to lobby for higher grades, while professors do not have an incentive to resist these demands. If a professor decides to lower the average grade, this will lead to poor evaluations, which at least theoretically could hurt the professor's career, and few professors are willing to take the risk.

And I totally agree with you about the reason grade inflation continues to happen. At first glance, it is good for the students, not bad or good for the professors, and good in the eyes of administration.

What I hope the country realizes in the very near future is that we've been steadily destroying the credibility of a US education for years now (at least at the undergraduate level and below; I couldn't even begin to comment on graduate schools in the US). This is obviously bad for everyone who was educated in the US (and I'm guessing a good thing for someone from say Europe wanting to get a job in the US). The issue isn't even whether we're putting out a better educated student on average than in years past in the US, because even if we assume for a moment that we are in fact putting out a better student, we still aren't doing as well as we should be relative to many other modern countries considering our resources.

I don't mean to sound doom and gloomy but I can't help but think this is an utterly gigantic issue due to the unthinkable runoff damage caused by being on a downhill slope from an education perspective in a world of new big shot growing countries whose education systems are on an upward slope.

[++$_]

However, I think you were implying that the professor make the class more rigorous which would most likely result in a lowered average GPA.

Yeah, that's what I meant.

What I hope the country realizes in the very near future is that we've been steadily destroying the credibility of a US education for years now (at least at the undergraduate level and below; I couldn't even begin to comment on graduate schools in the US). The issue isn't even whether we're putting out a better educated student on average than in years past in the US, because even if we assume for a moment that we are in fact putting out a better student, we still aren't doing as well as we should be relative to many other modern countries considering our resources.

I don't mean to sound doom and gloomy but I can't help but think this is an utterly gigantic issue due to the unthinkable runoff damage caused by being on a downhill slope from an education perspective in a world of new big shot growing countries whose education systems are on an upward slope.

Is it really that big a problem? I mean, there is nothing inherent and sacred about the letter C that indicates "average understanding." If it turns out that an A- indicates average understanding and that an A indicates good understanding, that's fine, just as long as everyone realizes that is the case. (And currently, most people do.)

There are many serious issues with the US system of education, which is massively failing in many ways, but the changing association of letters with performance levels is not one of them.

[Izawwlgood]

I feel there's serious grade inflation on all levels of the gamut. In graduate school, it's to their benefit to see us through, and they'll be pretty lenient to ensure we get there. In state schools, each student is a potential future donor or prestige puller for the university, and they should thus err on the side of helping them along than on the side of holding them back or having them drop out.

[ImTestingSleeping]

Is it really that big a problem? I mean, there is nothing inherent and sacred about the letter C that indicates "average understanding." If it turns out that an A- indicates average understanding and that an A indicates good understanding, that's fine, just as long as everyone realizes that is the case. (And currently, most people do.)

There are many serious issues with the US system of education, which is massively failing in many ways, but the changing association of letters with performance levels is not one of them.

I think it is a big problem. One reason being that grades are capped by some mark which is the highest, i.e. A, 4.0, etc. If it continues to inflate, the actual scale continues to get smaller (with the same capped mark) and it becomes increasingly more difficult to differentiate between students. B is the new C, A is the new A-, but what is A? A is just devalued. Students who would have received an A- before, are now receiving A's. What about those students who would've received A's before? They get SHAFTED.

Shafted in more than one way too, because at the same time, it becomes increasingly difficult as a teacher to teach the same level of material as years before while landing students in the new, higher mark range that makes the students, parents, the state (and therefore school administration) happy. What do they do to get kids in that new range is dumb down the classes.

Of course, there will always be the kids who just don't try and do poorly no matter what. What I think we'll begin seeing is a switch from this:



To this:

[Chen]

Well I don't see an issue with a class having an average of B+ or even an A. If the students can cover all the material properly they deserve their top marks. To do otherwise would require a shifting of grades so that the people who did the absolute best were the ones with As and the "average" would have Cs (i.e., curving). However, this creates problems comparing one class to another even if the material is the same because grade now becomes dependent on who else is in the class.

Grades are basically useful for grad school application and MAYBE your first job. I think saying this is a BIG problem is a tad alarmist, considering grades themselves are not terribly useful anyways.

[jmorgan3]

I agree that you guys are being a bit alarmist.

Technical fields tend to have lower grade inflation, since they have strict curriculum standards and professors who spend most of their time researching. In addition, hiring decisions are usually made on the basis of work/research experience in addition to GPA, so grade inflation matters even less.

People in non-technical fields usually don't get a job related to their major anyway, so their GPA isn't entirely relevant, either. Those that do related work often have graduate degrees, which have a non-inflatable research component.

[cjmcjmcjmcjm]

If I were grading, I'd make it so that there are "hard" standards for A's and F's (eg ≥92% is some type of A, no matter what; and <50% is a failure) and distribute D's, B's, and C's on a curve. Then again, my Diff Eq prof said that he's had semesters where no one deserved an A and semesters where everyone got A's and B's because they met the standards

[doogly]

Last exam the mean was 80 and the stdev was 10. It was so on point.

[Bakemaster]

If a bell curve is the goal, why is that? What's inherently virtuous about a bell curve? Yeah, we see it a ton in nature, but cool your jets, Rousseau; the natural is hardly the desirable if we're talking about higher education, the whole point of which is to elevate the student above the level of the natural.

Furthermore, if any particular mean or spread is upheld as some kind of goal, how do you assess an honors class? If your goal is 75%, middle-C, averagest of the average, should that be your goal in every class separately? In every semester separately? Over the lifetime of a particular course? Can one class be more capable that another, and if they are, should their average be higher?

In my opinion, one reason so-called "grade inflation" goes largely unchecked is because it's apparent to most people in positions of authority (over admissions, hiring, etc.) that grades are only one tiny portion of the story of a student's academic achievements. That's why we have to take standardized tests outside of class, obtain recommendations, complete instructor evaluations, write essays, and so on and so forth. You get grades during part of your life, and you do actual useful stuff (hopefully) during other parts of your life.

[Dopefish]

Last year I had a prof who felt the average should be 60 in an undergraduate class, and 70 for graduate classes. Also, no periodic tables since everyone in a chem major should have it memorised (this is a second year inorganic chem course, thus requiring knowledge of the positions of most of the transition metals, as opposed to the first dozen or so that most people actually know). I since built my schedule to never take any classes from him again (although I barely clawed my way to an A-).

This is in Canada though, so maybe folks like him need to migrate down south. Although I must say he was the exception, most of my classes have had 80ish averages, which is still high-ish compared to a proper average of 70...

I think I'm inclined to blame the internet to some degree. It's not necessarily the grades are being given away for less knowledge, it's that with the internet, many of us can quickly find the relevent information to learn from (or even just find similar enough examples to get through the homework for those extra points). I would imagine that would influence every sufficiently developed region though, not necessarily just the US. I don't know a solution for that though (or even if it's something that needs to be solved) since you can find virtually anything on the internet, and while asking those more qualitative deeper questions that would normally be used to seperate the A students from the rest may work, it doesn't seperate everyone else. Thus you end up with marks centered around 80 with a relatively small standard deviation, with most of the outliers on the 90's side, and it's difficult to pull the curve backwards, without doing stuff that would most likely be considered unfair by all students (i.e. giving a test/exam that realistically takes 2 hours to complete, in a 1 hour period).

[_Axle_]

My issue with this topic is . . . is there any evidence of this going on? Are there studies, reports, what not about grades being inflated?

Could they be going up for other reasons, like the kids actually are learning more than before? Maybe less people are taking the risk of college, so the dedicated are the ones still going, thus raising the average?

[doogly]

There are definitely studies that this happens.
At Dartmouth, where I did my undergrad, the solution was to list the median grade for every class (with >8 or so people) on your transcript. There was no suggestion that it is "wrong" for the median to be an A-, just that when this is the case, your B+ should not be interpreted the same way as a B+ in a class with a B median. Other colleges (eg, Wellesley) have more heavy handed solutions.

[achan1058]

You know, from playing video games, I have another suggestion. We should start having grade S, SS, and SSS.

[Dopefish]

I once got ~140% on a test that had an entire page of it made bonus, since we were told prior that that material wasn't going to be on the test and so we didn't have the appropriate formulas on our sheet. I ended up getting everything right anyway since first year stuff is easy and dimensional analysis works wonders, so the grade distribution for that single test was a nice little normal curve centered at around 80, with a lone outlier way up at 140. I enjoyed that thoroughly (and what's more, the same questions were repeated on the final exam). My overall mark in that course (first year physics) was 104 in the end, but unforunately they don't offer A+++'s (or SSS if you're into that).

[GenericAnimeBoy]

Spoiler'd for flagrant venting: I dislike bell curve grading.

Spoiler:

Seriously, FUCK bell curves, and FUCK you if you believe in bell curves as the ideal grade distribution.

Grades need to be an objective indicator of individual performance, not a comparison among students. I'm attending a public university, and I'm paying 2.2*10^2 metric fucktons of money ( in terms of value to me, anyway...to the university it's not much of anything) and placing myself substantially in debt for the opportunity to better myself and my value to society. It's reasonable to think *that* selection criterion in and of itself filters out a significant amount of the low achievers (not all of them, thanks partly to parents who force their kids into college, but that's an issue for another thread).

[doogly]

They're not ideal, they're the result of the central limit theorem. It's a math.

[GenericAnimeBoy]

They're not ideal, they're the result of the central limit theorem. It's a math.

The central limit theorem applies to random variables. Grades are not random by any definition of the term I've ever heard.

[ImTestingSleeping]

Spoiler'd for flagrant venting: I dislike bell curve grading.

Spoiler:

Seriously, FUCK bell curves, and FUCK you if you believe in bell curves as the ideal grade distribution.

Grades need to be an objective indicator of individual performance, not a comparison among students. I'm attending a public university, and I'm paying 2.2*10^2 metric fucktons of money ( in terms of value to me, anyway...to the university it's not much of anything) and placing myself substantially in debt for the opportunity to better myself and my value to society. It's reasonable to think *that* selection criterion in and of itself filters out a significant amount of the low achievers (not all of them, thanks partly to parents who force their kids into college, but that's an issue for another thread).

Grades are by their very nature a comparison to other students. I don't see how you could get around that.

If you're spending a lot of money to attend university, you shouldn't feel like you deserve good grades in return. They should be in the business of providing quality education at a higher level, not getting you out of their with a solid GPA. That's on you.

I understand everyone's points on the bellcurve's not being optimal for a grade distribution, but to me it seems that the pool of talent in every class and school works on a bellcurve. I would think the classes should be difficult enough that something close to a bellcurve shows up in the grades.

I like the median of every class being post on the transcript as mentioned earlier. That would honestly be enough to make me happy on the issue.

[Yakk]

They're not ideal, they're the result of the central limit theorem. It's a math.

The central limit theorem applies to random variables. Grades are not random by any definition of the term I've ever heard.

A measure of norm 1?

[++$_]

They're not ideal, they're the result of the central limit theorem. It's a math.

But the grade of a random student may not be representable as the sum of independent random variables.

[achan1058]

They're not ideal, they're the result of the central limit theorem. It's a math.

The central limit theorem applies to random variables. Grades are not random by any definition of the term I've ever heard.

But the grades of students (or rather, whether a student knows X on a test) are effectively random variables. In a way, be glad prof "bell curve" (from what I heard/seen, they usually simply translate the grades with a linear function, the shape of the bell curve comes naturally from the large number of students). Otherwise, you will see significantly higher fail rates. (in STEM at least, it's very rare that profs scale down, but they usually scale up, to possibly 40% at a time)

Edit: I will add that scaling grades might not necessary be good for a small class (unless the grades came up to be horrible because you set your exam way too hard), but for large classes, the chance that everyone is a genius (or an idiot) is so slim that that the procedure is justified.

[Magnanimous]

Aren't bell curves just for really large classes, like over sixty students? I can agree with curving for smaller classes, but not to such a strict graph.

[doogly]

I've also never seen people force grades into a bell shape. When folks say the grades are curved, it is really just the linear shift, as mentioned above. F'rexample, I get two-bump distributions pretty often. I would never push it into a different shape.

[GenericAnimeBoy]

I've also never seen people force grades into a bell shape. When folks say the grades are curved, it is really just the linear shift, as mentioned above. F'rexample, I get two-bump distributions pretty often. I would never push it into a different shape.

I have an ECON professor who assigns letter grades according to the bell curve. He also treats trickle-down economics like the Gospel.

[doogly]

Ah, here's your problem! You should be hitting him with a stick!

[Bakemaster]

No, you should be urinating on him.

[cjmcjmcjmcjm]

I got an A- in one of my class because I didn't suck as bad as everyone else (class of 12). Truth be told, I would have gotten a B- or C in a linearly graded class. The prof also kept threatening to get rid of the curve and fail over half the class (required to graduate with degrees in Chemistry or Environmental Science) because we were performing so poorly when compared with previous semesters

[gmalivuk]

They're not ideal, they're the result of the central limit theorem. It's a math.

But the grade of a random student may not be representable as the sum of independent random variables.

Actually they probably can, if there's not too much overlap in the material covered in successive grades.

[++$_]

Actually they probably can, if there's not too much overlap in the material covered in successive grades.

Even if different material is covered, it often requires the understanding of previous material, or at least the same skill set that was required to understand previous material.

For example, my experience is that people who struggle to understand stoichiometry will go on to struggle to understand chemical equilibria, because they both require the same skill (the ability to apply math to a real-world situation).

Or, in an English class, if you get a poor grade on the first essay, it might be because you just didn't grok that first book, but it's more likely that it's because you weren't a great writer, in which case you still won't be a great writer on the second essay.

[achan1058]

Actually they probably can, if there's not too much overlap in the material covered in successive grades.

Even if different material is covered, it often requires the understanding of previous material, or at least the same skill set that was required to understand previous material.

For example, my experience is that people who struggle to understand stoichiometry will go on to struggle to understand chemical equilibria, because they both require the same skill (the ability to apply math to a real-world situation).

Or, in an English class, if you get a poor grade on the first essay, it might be because you just didn't grok that first book, but it's more likely that it's because you weren't a great writer, in which case you still won't be a great writer on the second essay.

Even if the variables are correlated. If you add enough together, you will get something that's approximately normal.

[Dopefish]

CLT may well apply, but it rather requires large numbers doesn't it? Even in a class of 100 students, that's not 'that' large, and many classes are smaller (not to mention in smaller classes, instances groups of friends working together so that if one gets it, they all get it make up a larger portion of the class, increasing correlation between individuals).

While the average over a whole series of semesters of a given course would probably be normal by CLT, I don't necessarily think any individual class would necessarily be normally distributed. Of course theres plenty of empirical evidence to support that they are, bimodal distributions are fairly common too, where you have those who get it and those who don't, so 'forcing' marks into a certain shape seems like it'd only cloud things further since it'd potentially cause people who don't understand to get the same grade as someone who did.