I generally saw a modified 100 percent absolute scale for more advanced upper-division classes where the threshold for “learned enough to pass” is significantly lower than “learned enough to get an A.”
As long as the tests are designed fairly (e.g. a small mistake leads to a small deduction in points, not a massive one) then that is reasonable.
One of my colleagues asked at a faculty meeting, “Do you see yourself as a weeder or as a gardener?” It was a rhetorical question that went without an answer, but later I sent her an email with the result of polling the faculty (made up as a joke, I emphasize!) : Weeder. Weeder. Weeder! Weeder, definitely! and What is meant by this question?
There is some of that attitude in the department, but there are also serious attempts to counter it.
Reading this discussion has been interesting, but much of it is like listening to a bunch of 13 year old boys talking about sex – full of enthusiasm and strong opinions, sprinkled with facts gleaned from Playboy, but it’s not based on any actual experience.
The best grading policy varies depending on the class that’s being taught, the norms in the field, and the level of the class and the students. Still, many professors in STEM fields do tend to give exams that have median grades around 50% or so. It’s not because they’re mean or “old-school” or delight in failing students. In fact, many will gladly curve an upper division class so that a 50% is a B+/A-. Here are some reasons why this system seems to work best (this perspective is based on teaching in a STEM-ish field in a very highly ranked department; not all of this applies to other situations) –
And yes, the range of abilities in a class can be enormous. You want to have problems that challenge the top students or even a future leader of the field; however, to a below-average student these problems might as well be written in Sanskrit. So you need to use the full range from 0% to 100% to cover this wide range of abilities. Anything else is just a bad test.
Nobody said challenging students to reach their full potential would be an easy process. But learning is a worthwhile goal.
I guess if their egos are bruised by only getting a 50% (which still might be an A-), then we could do what the SATs do and just given them 25% of the points (200 out of 800) for signing their name. If that’s what makes them happy, then it’s fine by me.
But this just isn’t appropriate for testing higher order cognitive skills. In some fields, it’s pretty much impossible to write a test for an advanced class where the average is a 90% unless much of the exam is so pathetically easy that it doesn’t test anything. The goal is to test students by giving them real problems that i) involve applying the concepts to new situations, combining multiple insights, or require a small creative leap ii) aren’t similar to anything they’ve seen before iii) can still be done by a human being in under an hour.
Just reading and thinking about one such problem can take a test-taker 15 minutes, so at most you can give 4 problems in a 3 hour exam period. It’s very hard to invent these problems … in a few cases, these would have been publishable results a generation or two ago. So it’s no surprise that you can’t expect the average student to get much higher than a 50%. It would be like demanding that everyone “master” baseball by being a .900 hitter. That would be ridiculous because hitting a 90 mph ball is just hard.
In some fields, the problems worth doin’ are just plain ol’ hard.
That’s one reason why people use relative curves. Most professors are happy to vary the curve if a class is unusually talented, but honestly in a class of 100-200 people the average ability level pretty much tends to be the same year to year.
Remember also that, unlike high school teachers, many professors don’t teach the same class year after year. Also, some material in really advanced classes is only a few years old and is often synthesized on the fly from academic journals by the professor. How do you think textbooks are born?
Here are two iron law of economics: 1) there ain’t no such thing as a free lunch 2) if it were easy, then everyone would do it. If you want to graduate into a job that pays $80K/year, then you’d better acquire skills that are worth that much.
Of course, we can devalue a college degree in an engineering field by giving almost all incoming freshmen students A’s or B’s too. By the end, we’ll have reduced a college degree to a high school degree. Then mommy and daddy can really complain that they’ve spent $250K sending their kid to college and the only job they can get is as a barista at Starbucks.
@al2simon Thank you for taking the time to type out that reply. I am grateful for professors who do want to challenge their top students.
I don’t think you really realize how dismissive your post comes off as, but you essentially lost credibility by opening with a character attack on anyone you disagree with.
It can be done and it is done at all stages in a person’s professional life. A university is probably the earliest place where you can start identifying high level talent, but it most certainly isn’t the last or the only one. Nor is it a prerequisite for high intelligence to be admitted to a top university (on factors that have a lot more to do with the parents than the student) and have the money to pay for it (likewise).
It sure as hell isn’t exclusive to university. And it sure as hell does require a solid foundation in the basics that you learn in high school.
And it most certainly isn’t true that you should put arbitrary “normal distribution” cutoffs on who is qualified to “advance the frontiers of human knowledge” because success in science is as much a function of luck (right place, right time, right approach) as it is a function of talent.
“We only care about the best of the best. All you other plebs can go on your merry way and have a nice insignificant life.”
Tests are designed to be as hard as they need to be. Writing a test on a standard letter grade scale is appropriate for most topics when what’s being measured is a rudimentary understanding of the fundamentals. For classes with more depth, a modified absolute scale is fair.
Occasionally it’s fair to adjust the scale because a test ended up being harder than you bargained for (having written exams before, I can see how this happens). If this happens on a regular basis then you are a terrible teacher.
It would be politically irresponsible for top schools not deny actual high-quality students the GPA hard requirement they need to get into med school or law school etc. Not as true in STEM, where advancement follows the principle of “I know a guy who knows a guy” (a professor our research group works with recommends one of his undergrads).
…
I attend Cal, and our STEM classes are typically curved to a B- average. The average on our exams tend to be pretty low like sub 50%. However, most students still get a C or higher and pass the class. Most engineers at my school graduate in 4 years or less.
More than of Stevens students are in the Co-op program, which requires five years due to the two working semesters (which are distributed with the regular on-campus residency) therefore the 4 year graduation rate is low and the 5 year rate high. Stevens has no part time undergraduate programs.
Stevens is an extremely rigorous school. Graduation rates by themselves do not tell the whole story.
That is, more than half are in the co-op program.
https://schedulebuilder.berkeley.edu/explore/ indicates a lot of B and B+ average course grades, including in courses commonly taken by engineering majors. However, it does look like some chemistry (1A, 1B, 3A, 3B), biology (1A, 1B), and molecular and cell biology (102) courses popular with pre-meds tend toward B- averages.
@intparent “This is pretty deeply ingrained in women, to blame themselves and assume they aren’t good at something.”
Yes. Yes. Yes. It seems to help that to talk a lot about effort, and very little about ability and grades. When they do their best, the grades will take care of themselves. We try to disconnect grades from ability.
@intparent “The bias in teachers honestly runs pretty deep, too.”
Sadly, they can’t see it right in front of them, and the parents often have the same biases.
Last week I heard a mom say, “My daughter just isn’t good at math, I was bad at it too.” I smiled and nodded, but I was thinking that with the right instruction and mental attitude, they both could have been great at math.
The problem with grading on a curve exists only where you have professors who feel that they NEED to fill out the bottom and top of the bell curve, whether anyone deserves to be there or not. If you happen to have a class of high achieving students, everyone should pass. If you happen to have a class where no one tries (don’t think this happens IRL), then there should be no As’. I think applying a bell curve arbitrarily to a class of <40 students makes no sense.
In the case of “1 question A students can answer, 1 that B students can answer, and 1 that C students can answer” - there is no problem at all, as long as those that are able to answer the C question actually receive Cs, rather than D/F because someone needs to fill out the bottom of the curve.
Yes, the example was with absolute grading, not grading on a curve. However, then you may have students getting 30% for a C and 60% for a B. The raw percentages look bad (which previous posts have criticized as discouraging students), even though they are not really bad.
When Mr. Fang took the Putnam, he only got 5 of the questions right-- he says there were 15 questions his year.
He was 15th best in the US that year.
Real tests that test have hard problems on them.
Every undergrad STEM test should not be modeled after the Putnam.
A test meant to separate the super-super-elite from the super-elite and merely elite at math would have much more difficult problems than a test for a typical college math course meant to distinguish between really good (including but not limited to the elite at math contest takers) A students, good B students, solidly passing C students, and barely/non-passing D/F students.
@“Cardinal Fang” Don’t you fret on other threads about kids throwing themselves in front of trains?
I don’t say that an exam for a freshman should be as hard as the Putnam, but I do say it should have some problems that are hard for the students in the class. The student should have to think about them, or at least some of them.
I personally don’t believe that exams are the best way to give students a set of problems that advanced for the field or open research problems. It’s easier to think clearly about difficult problems when the stakes are lower and you have plenty of time to concentrate on solving them.
Most people don’t become brilliant when you toss needless stress their way.
Part of what you get when giving more difficult problems is whether students can even begin to have a thought about them.
I often give problems on exams or homework that some students don’t even attempt to do. Those who tried but got it wrong at least still have brainwaves,
Going to be honest – I don’t think an exam is the place to give kids material they have not been taught (or permutations they have not seen in homework or lab assignments) and expect them to figure it out. You aren’t teaching the material iIf that is how you write a test.