A classic curve is normative: you decide how many As, B, etc you want and then draw the cut scores based on that.
Scaling grades is much more common. This is when you decide a certain level of performance represents A, B, C work, and so you adjust whatever % of the points that is to the 90% we’ve decided A should be.
For some reason, people think a scaled test is “easier” than a non-scaled test. But really, scaling saves you from havibg to pad out out test with questions you know they all know just so even the people who barely know enough to pass can get 70%.
Does this mean what was “B” work 10 years ago can now be “A” work because the goal posts are moving? Or are there hard lines that demand a level of knowledge (like if you were taking a test to be a surgeon or pilot where you never want to see a drift down in ability/knowledge)?
It means I had a chem teacher in high school who graded on a class average, so the lowest grade was always an F, even if it was an 85%, a grade the would be a C+ or B- from most teachers.
Guess who generally scored at the bottom end of the scale?
I dropped the class the last day to drop with a plain “W.” The next day, you got a “W-P,” or “W-F.” Withdraw-passing, or Withdraw-failing.
Possibly, because the work is being compared to the fellow students you are being graded with.
The first year of law school is pretty rough because there is almost always a pretty harsh curve for core classes. I think accreditation might depend on it (didn’t look it up). So there might be only so many A’s and there will be the same number of guaranteed fails. And your scholarship depends on your GPA being above a certain number…
I’m curious whether @Fretful_Porpentine still stands by this comment from that old thread linked to in the OP:
An experienced teacher probably has a much better idea of what constitutes exceptional or adequate work, at the level of a particular class, without having to judge by comparison with what the rest of the students in that class did, which is what grading on a curve does.
Good Lord, that’s a really old thread! As in, I was teaching my first class ever when I posted that.
I suppose so, but … doesn’t that just mean grading on a bigger curve, one that judges students by comparison with all the other students you’ve ever had at that level? And I think that is, mostly, what I’ve always done and still do.
I mean, it depends on the course, the institution, and the instructor.
But you can have a very rigorous test where, say, 50% correct is A work. The AP Physics C test is like this. Its just very efficient and doesnt bother asking any easy questions. Covers a semester of material in 90 minutes, becauae everything is so interconnected. If a student can answer a question on topic D, you can assume they k ow topics A, B, and C, so no need to have any queations on them.
On the otherhand, the AP government test requires like over 80% of available points to pass and over 90% to get a 5. Its also twice as long. There, knowledge of one topic doesn’t mean ypu know others, so the test needs a bigger sample.
AP cut scores are set by having college professors insert AP questions into their own assessments and comparing student performance on those items to student performance in the course. And then the final cut points are aet my a committee of professors looking at that data.
Do the people teaching the course need to submit their grading criteria (there is a curve, this is how it works, or no curve at all) to the school/college they are teaching in? Or is it all up to the teacher/professor?
(I suspect the answer is, “It depends where you are.” No hard and fast rules. To which I would ask are there teaching guidelines on this stuff that educators should try to adhere to?)
One thing to remember is that students see assigning precise grades as a very central part of the instructor’s job, but from the other side of the desk, it really isn’t. My job is to teach, not certify. As a secondary teacher, I literally got no training in grades. We have a district grading policy, but it is pretty meaningless.
There are obviously some grading standards. For example, the department would probably not like it if you kept failing three-quarters of the students in every class.
AFAIK @MandaJo is correct: you can set an easy exam or a really rigorous one. Now, you surely have some idea which are your best students and therefore of an “A” performance. Somewhere below that will be a “B” cluster, and so on. The “curve” comes into it when you make a final decision what the actual cutoff is going to be and decide, for this exam, 38 points gets one an “A” but 37 is only a “B”, and so on. Presumably you will have done this so no, or at most very few students, got exactly 37 points because they will all end up in your office arguing for 1 or 2 extra points.
ETA To be clear, I am not talking about standardized tests.
I took an upper level physics course where we were lucky to score in the 70s and 80s on our tests, and the professor graded it in such a way that one guy got an A, a bunch of us got Bs, and a bunch of people got C’s. I don’t know that there’s any name for that curve, other than it kind of sucked.
I deliberately kept it simple, but should have said, some places have more fine-grained systems which let you assign the one or two really outstanding students an “A+”. That way a bunch of students can still get an A and even some an A−, you don’t have to go straight from a single A to a bunch of B’s.
I know of someone who was really pissed off that, on an important final exam in Germany, received a 1.2 while his friend got 1.0 (a perfect score). I guess for more than the maximum possible they could have gone to marks below 1.0, but that is not how they rolled in that school.
I mean you see (imagine looking at a histogram of all the scores) a cluster at the top and (presumably) they will all get an A. I never said it should be based on one particular student, only that there may be (on a difficult exam) a freak outlier, but you can just give that person (or more, this is not and, if you are a good teacher, should not, be a “there can be only one” situation) an A+.
Maybe if one is an outstanding teacher in theory everyone could wind up with an A? However unless the class is tiny there may be some students who did not really study or work as hard, so it is unlikely. The students themselves (and the administration) would be unnerved if everybody got an A. I am talking about big undergraduate classes, not a graduate-level seminar or something.
I took a *beginning chemistry course in college. It was taught at way too high a level and at the end of the semester a 56 out of 100 on tests was an A. I got a 26 out of 100 on one test. That was a C. I don’t think I ever scored above a 30. I got a C in the class, and came out of it feeling stupid beyond belief.
It is kinda weird that this aspect of testing is being ignored in this thread so far. I get it is a different thing and maybe no one here has expertise with how they are made. But they loom so large in education in the US these days they are also hard to ignore.
Are there curves for kids in the inner city who have a provably lesser education than a kid in a wealthy suburb/school district? Or are they all on the same grading curve for a standardized test?
Maybe that would move this topic to GD or IMHO (fine by me if the mods want to).
I would add to previous replies that rigor in grading is generally reinforced by feedback from instructors farther along in the students’ curricular trajectory.
That is, if you teach a course that is a prerequisite for other courses and you’re letting a lot of really underprepared students through with a passing grade, the instructors of those other courses are going to complain that their students haven’t learned the necessary background material, and you will hear about it.
Personally, I grade individual answers based on a sort of “mastery” rubric: did you display enough understanding, knowledge and effort to get you at least halfway to a valid answer? most of the way? correct except for some trivial mistakes? etc.
If you averaged only about “half mastery” of the material over the whole assessment, I usually consider that a C. “Full mastery, maybe with some trivial mistakes and at most one more serious error” is what it takes to get an A. (If on a scale where A+ is allowed, you won’t get that with even one major mistake.)
A grade of B lands about halfway between those two cutoffs, usually around the score that’s the median/mode average of the class performance.
“Secular” changes in class performance over the course of years or decades, due to changing levels of student competence, are not reflected in the grading “curve”; they’re reflected in the course and assessment design. (I no longer require students to perform any but the most trivial arithmetic computations in a problem on a no-calculators test, for example. I personally may think it’s better for students’ deeper understanding if they have better facility at calculating without digital assistance, but that’s not what they’re in this course to learn. And to be fair they really have very little need to do basic arithmetic without a calculating device in any real-world circumstances.)
I wrote a pretty detailed comparison about how AP Physics C and AP Government abd Politics are scaled, and why the scale is so different, and why that really doesnt say anything about the relative ease of the course or the test.
I’m not sure how that’s ignoring standardized tests.