In your opinion, what constitutes “grading on a curve”?
For example, if a teacher applies a scale where 90-100%=A, 80-89%=B, 70-79%=C, 60-69%=D, and 59 and below=F - did the teacher apply a curve?
What if the teacher above finds his class did poorly, so he drops the standard for each grade 10 points such that 80-100%=A, 70-79=B, etc. - have they now curved the results?
How about, instead of lowering the scale 10 points, the teacher adds 10 points to each students’ score?
Or does a curve require that a teacher give As to the highest 10%, Bs to the next 20%, and so on?
Just curious as to folks’ opinions. I was speaking with a teacher who said they had applied a curve in the second example above - where they adjusted the 90-80-70-60% scale down 10 points. I said that wasn’t what I considered applying a curve. But I have no knowledge of how that term is approrpaietly used in educational circles.
No. And I won’t hear anything from any of you heretics who think there are grading scales where 90% isn’t an A. The 90-100 = A et cetera scale was written on the back of the Ten Commandments. Written under it is “Thou shalt never curve grades down”.
I’ve never heard of a teacher doing that.
Something like this is usually what is intended when they drop the cutoff for an A. The teacher has an idea of roughly how many people in the class should be getting each grade, and adjusts the grade cutoffs to get approximately that result.
The sloppy way to do it is just to take the highest score and make that the new 100% for everyone. That isn’t a real curve though. Truly grading on a curve means that that that the mean grade is a ‘C’ and there are as many ‘F’'s as 'A’s no matter how well the class did overall. I had a few of those in college and they can be difficult and tend not to be very popular. That technique is sometimes used on psych tests to help illustrate a point. You actually had to break out a z-score calculator to figure out your real grade.
In my college courses, we were graded on a bell curve (which is what I think you mean by “curve”). So once all the tests were graded, the grades (A, B, C, etc) were relative to how the rest of the class did. I don’t know the exact math (although I’m sure you could easily find something online) but the end result was that the kids with the highest scores relative to the rest of the class were the ones who gots A’s. I think this fits best with your example that X percentile gets an A, the next highest percentile gets a B and so on. So even if I only got 60% of the answers correct, I might still have gotten an A because everyone else only got 40% of the answers correct.
In fact, this is basically what happened to me when I got my first college chemistry test back with a 63. I was all prepared to call my parents and tell them that I failed my first test but it turns out, that was a B. I love that bell curve!!
Technically, grading on a curve implies some sort of non-linear conversion of percentage correct to grades–though it doesn’t have to be to a bell curve (your example). For example, some teachers use a “square root curve”, where you take the square root of the percentage correct and times that times 10–under this curve, 50% correct becomes a 70 on a 1-100 scale (or a C on a letter scale, which are unusual in high schools these days) and 90% correct becomes a 95 on the same scale.
Colloquially, any shift in grades is called a curve.
To me it means that the number of students getting As, Bs, etc, is decided before grading starts. So the top X% gets an A, the next Y% gets a B, etc. You assign letter grades to match a certain distribution, not necessarily bell-curve but AIUI that’s where the name comes from.
Just moving the cutoff % for an A up or down based on how hard the test is? I wouldn’t quite call that a curve, since you’re not necessarily competing against other students to get one of the limited number of A’s and B’s.
As I understand it, this is the “traditional” method of curving: mapping the scores to a literal bell curve and assiging the final grades based on the results. I have heard anecdotal accounts that there are still teachers who use this method, but I’ve never actually encountered one.
In my own experience, to a man (or woman), every teacher I’ve ever had who graded on a “curve” used the same method: subtract the highest score in the class from 100, add the difference to every student’s score, and assign final grades based on that, just as Shagnasty describes. This held true in both junior high and high school (class of '02). To my knowledge, none of my college classes have been graded on a curve.
Throughout my entire schooling through law school I never encountered a true bell curve. My understanding is that a bell curve would require that as many students get Fs as get As, and the same number get Ds as Bs. I’ve never been in a class where as many folk got Ds and Fs as got As and Bs. Usually the curves would reflect something like top 20% get As, next 30% get Bs, everyone else gets Cs (except for a couple of folk who work really hard at earining a D or F).
Anne: The method where the teacher adds points to everyone’s scores is common in my kids’ high school. So my kid will get a 103%, even tho she got some questions wrong. For whatever reason the teacher wanted to stick with the 90/80/70 scale, so she added a certain number of points to everyone’s scores to get a “more desireable” distribution of grades within that scale.
My question with that method was, souldn’t the added points reflect the number of questions answered correctly? I’m no mathmetician, but it doesn’t seem quite right to me thatboth the kid scoring 40 and the kid scoring 90 get the same number of points added to their score.
Manda and sugar - I think you hit the nail on the head as to the disjumct between me and the teacher. I was speaking as to the “techinical” definition of a curve, whereas she was using the far more common colloquial def.
Letter grades on report cards are pretty unusual, in my experience: report cads say “89” or “94” or “72”, not “A” or “B” or “C”. There may be a district policy that 90-100 is an A or whatever for the purposes of GPA, but it’s a 100 point scale. In cases like this, a teacher has to add points to a grade, not redefine the letters, because they don’t turn in letters to be printed–they turn in a number between 1-100.
In high school, and I think a few timeis during college, whenever the teacher would grade with a curve he’d get the highest grade, figure out how many points the student would need to bump that highest grade to 100% then give those points to everyone else.
So if the test had 100 one point questions and the highest grade was a 95, everyone else would get an additional 5 points.
Seem pretty stupid to me but I’d never tell a teacher that
Let’s say I had a class of 25 students. I would mark the tests and then write the marks down on a piece of paper. Then I would look for natural groupings. If there were five students with marks around 70 and nothing higher, then they would get A. If there were four and one around 90, that was a problem and it would depend on whether I thought that one was unusual. And I would continue in that fashion. It was also possible that there be no A if I felt the class was unusually dense. Much the same at the bottom. Students way below the median would fail. If there was a large cluster at the bottom and no one way below, they would all get the same mark, but whether D or F would depend on whether I felt they deserved to pass in some platonic sense.
The province of Quebec used an interesting way to curve marks. They would average class marks with the provincial exam marks. But first they would adjust each class’s class marks so that the the class marks had the same mean and the same standard deviation from what that class got on the provincial exam. And this adjusted average was the final mark. The effect was to take into account the class work in such a way that an unusually hard or easy teacher could not skew the results.
I had a Statistics professor in school who got a double bell curve. He gave the students at the top of the curve all As (they’d scored 100), then the next curve got Cs, Ds and Fs. Each of his tests had ten questions and gave no partial credit.
He was an elitist jerk who didn’t realize that a core group of students, almost all of whom were pre-Med, got a hold of his old tests and were cheating. He was possibly the worst teacher I had in college, taught the class by reading from the text and spent the rest of the time complaining that it wasn’t fair that he wasn’t teaching in Cambridge (Mass) and suggesting that anyone attending our school was an idiot.
I later taught myself a sufficient amount of statistics by reading someone else’s book.
This is what a number of my professors in college did. You can’t really fit a bell curve to the kind of grade distributions you’re going to get with 10 students in the class (which is not unusual for an upper-division course for physics majors), and the raw scores are likely to be low enough in those classes that some kind of curve is needed so that everybody doesn’t flunk.
I only had a few courses in my life that were graded on a curve and they were all done basically like this. Unlike others here, the only place I ever got graded on a curve was in college. Heck, it was usually the first question some kid would ask in each class, each semester: Do you grade on a curve? This technique can lead to some really interesting results and very unhappy classes, like when one person gets an A and the next highest grade is a D.