Brigham Young University Home

Why do some professors grade on a curve?

Dear Professor,

Why are so many classes at BYU “curved”? It seems like in many classes students aren’t encouraged to help each other or work in groups, but compete and hope for other students to fail the test. I know the “real world” is competitive, but should education be as well? Isn’t the whole (working together as a group) greater than the sum of its parts?

Dear Student,

Thank you for the question! The issue of grading and fairness hits home for both instructors and students. Accurately assessing how much students have learned is one of the most challenging aspects of teaching, and assigning a fair grade isn’t any easier. As one of my colleagues pointed out, three purposes for assessments (and for eventually assigning grades) come to mind:

  1. To discern whether students have achieved course goals for learning;
  2. To identify content areas where students are performing well or poorly; and
  3. To improve learning activities and assessment practices within the course.


Instructors use different strategies to accomplish those goals, which is where two main types of assessments come in. We can either compare each student’s performance to a set standard of performance (criterion-referenced assessment), or we can compare students’ performance to that of each other (norm-referenced assessment). The latter option is where the so-called “curve” comes in.

So why would an instructor use a curve to assign course grades, rather than simply compare students’ performance to a defined standard of perfection and let the scores fall where they may? Here are the reasons another colleague cited for using a curve:

  1. Some faculty who grade on a curve believe that students will be motivated to study if they know that only a limited number of students can receive an “A” grade.
  2. Some who grade on a curve believe that it is the only way to guard against grade inflation.
  3. Most who grade on a curve do it because their discipline has a long-established tradition of such a practice.
  4. Some who grade on a curve do so because they are following a high-stakes testing model such as the ACT or the GRE. These tests are used to determine which students are accepted or rejected by educational institutions. Some faculty believe that the only way to prepare students for such norm-referenced tests is to give such tests in their own courses.

You’ve probably had some experience with curved grading as a means of preparing students for national norm-referenced tests or for selecting students to participate in certain areas of study. The purpose of such curves isn’t primarily to find out what students know, but rather to sort students into categories for decision makers to use when admitting candidates into competitive fields or programs. As one instructor explained,

Classes that grade on strict curves are (in my experience) “weeder” classes. Some majors and professions—especially in scientific and technical areas—sound appealing but they require a very good grounding in math and science. The people who succeed in these weeder courses are both talented and hard workers—those who are willing and able to do what it takes to reach the top. The class is designed to single out those students who have the potential to become the best in their field.

Getting discouraged? Don’t! In some cases, curves actually help students. For example, one instructor noted,

Every time I have curved the grades in a challenging class, the curve has helped the students by bringing up everyone’s score. I think the only time a student would be harmed by a curve is if the course is easy. I can envision a scenario in that case when students’ grades would be forced downward, but I’m not sure this happens very often. But in ten years of teaching, I can honestly say that I have never once decreased a student’s grade by applying a curve. That’s because in every class I’ve taught (~90 classes), students have fallen out along a curve without my trying to make it happen! Such is the magic of the normal distribution.

Another instructor laid out her typical strategy that combines criterion-referenced assessment and slightly curved grading to the students’ advantage:

Some faculty, myself included, usually grade by a fairly objective standard; i.e., 93–100 percent=A, 90–92.9=A-, and so on. However, when calculating final semester grades, I notice that total scores are often distributed in clear cohorts. As a result, I may shift grading standards slightly either upward or downward by a percentage point or some fraction thereof, depending on how many scores appear in a particular “lump” or grouping. The real-world factor thus enters in to an extent, but arguably in a natural, not excessively punitive, way. (In other words, a typical “A” student doesn’t become a “B-” student as a result of this approach.) Incidentally, my courses are heavily team driven, with an emphasis on cooperation and peer review.

Finally, it’s important to remember that you can influence the course of your own education. If you have questions about grading or assessment practices, or if you notice that competition among students is lessening rather than building opportunities for collaboration, you can talk with your instructor about your concerns. Approach him or her in a spirit of respect and sincere inquiry, and find out the rationale behind the assessment/grading scheme. Be ready to listen—and be ready to offer workable alternatives backed up with sound examples and evidence.

Even though it may feel as though instructors and students sometimes are on opposite sides, we really are all in this together—we all want the best possible learning experience. One last contribution from another instructor sums up the situation very well:

Traditions, such as grading on a curve, are changed when many begin to question them and to suggest viable alternatives. Grading rubrics, like any pedagogical practice in a classroom, can be improved by students and faculty working together to find more satisfying and effective ways to increase learning.

Comments

Haley &mdash Jul 29, 2009 @ 3:15 pm

I was wondering if a Pass/Fail class will affect my GPA at all?

Rodrigo &mdash May 15, 2011 @ 8:48 pm

I think the answer is not referring to the question. Grading on a curve, in all classes, makes top students to study by themselves and not trying to cooperate with other students because they know is not good for them. Its a selfish system, just ask studens how they behave when professors curved. Wouldn’t be a better system to curve from the other side? This means, when the whole class makes it better, everybody has better grades? This could make students to help each other and act as a group, they all learn more and everybody benefits from it and goes happy back home. Could it be possible? I don’t know mathemathically how to do it, but there must be a way.
Greetings

Nathan &mdash Dec 7, 2011 @ 2:43 am

If a professor grades on a curve it means their class objectives are not at an appropriate level. Let’s remember that an A means perfect. Few students should be getting A’s. If too many get A’s, then course objectives are too easy. A nice curve means that objectives are appropriate. Applying a curve where one does not naturally fall is artificial and bad teaching practice. It’s like adding red food coloring to bad meat so it “looks the way it’s supposed to”. Besides, why should one student’s score depend on another student’s performance?

Ohioprofessor &mdash Dec 15, 2011 @ 5:49 pm

If I could ask the same exam questions every semester I could evaluate every student in a consistent fashion and make valid comparison’s from semester to semester. Unfortunately exams can’t be reused for reason’s that should be obvious (I hope). I grade on a curve because it is impossible to consistently come up with several hundred exam questions every year that equally and fairly measure the same material. Why should the class of 2011 get easier exams and better grades than the class of 2010 just because I had difficulty writing questions that year. If I assume that the median grade out of 200 students each semester will be the same if they are given the same questions, then that leaves me free to write difficult exams knowing the curve will bring everybody up to the grade they deserve. The assumption that the median student will be of similar capabilities year after year is valid I believe. And by giving difficult exams I can challenge (and identify) the occasional genius. Can’t do that when half the class gets 100%.

Martin &mdash Mar 30, 2012 @ 3:23 pm

Ohioprofessor has it quite right. Adjusting grades controls for terms when instruction and assessment is better or worse than in other terms. As with human activity, teaching is not exactly replicable each time a course is taught.

Calprofessor &mdash Jul 20, 2012 @ 12:10 pm

On the other hand, curving, as described here, does not control for variability in student performance. Why should a student who performed better than all the A students in previous years deserve a B because they were unlucky to take the class with 10 super motivated students?

Curving does address grade inflation if the proportions assigned each letter grade are kept constant. Curving DOES NOT address performance deflation, which is more troubling.

I prefer to assign grades based on criteria associated with the course goals.

IDProfessor &mdash Jun 11, 2013 @ 11:00 am

I think Cal professor has it right on. A studious look at curve grading will reveal that it doesn’t do most of what teachers would like it to. In response to the four reasons listed above:
1. Motivation. Grades are intended to be evaluations of performance that incidentally have motivational consequences. When they are turned into tools of motivation that incidentally have evaluation properties, the purpose of the grade has been undermined. Also, if a student is motivated by getting a grade, then a curved grade motivates the student to do better than others whereas a criterion-referenced grade would motivate the student to master the material.

2. Grade inflation. As Calprofessor pointed out curve grading doesn’t address performance deflation. Since grade inflation is defined by a rise in grades without a corresponding rise in performance. Since student grades would have the same distribution, if there were to be a drop in student performance, then curve grading would actually cause grade inflation.

3. Long established tradition. World was believed to be flat for a time . . . need I say more? Simply not sufficient reasoning and needs more substantiation.

4. Preparing folks for standardized tests would be a great reason to do curved grading, if the professor could show that learning to deal with the ranking of the students in a predetermined distribution is an actual objective of the course.

Curve grading has the purpose of differentiating and ranking students. For the courses where the ranking is more important than evaluation, then it should be done with that reasoning. For courses where that is not the case, the other reasons for using it are simply misinformed.

Chua &mdash Oct 28, 2013 @ 5:46 pm

IDProfessor has got the point clearly, especially in the last statement.
Standardized tests that cater to a large number of students for ranking purposes would do well to use curve grading as it can be assumed that the yearly median of a large number of students is the same year to year.

For smaller institute based classes with much fewer students in which courses are largely influenced by previaling teaching and learning conditions, outcome based learning with assessment tailored to assess learning outcomes based on the objectives of the course are more relevant and beneficial.
By focussing on the attaining of learning objectives, the overall student cohort of the institution or class may be enhanced and thus have a better outlook when and if they were to undergo standardized tests for further study further on.

Jo &mdash Dec 5, 2013 @ 3:09 pm

Often times grading on a curve brings grades up. I do agree that it is a good system to weed out individuals who are not prepared for higher level courses, however I think it shows a HUGE flaw in teaching style. If your students are doing so poorly that you have to grade on a curve just to bring grades up, obviously you’re not teaching the subject material very well.

TXPhys &mdash Jan 10, 2014 @ 2:19 pm

Take another theory of grading — the one implied by its name. I am evaluating the capacities of a bunch of college students, then passing that evaluation along to other professors, employers, the students themselves, etc. I am measuring them and reporting on that measurement — learning speed, comprehension, diligence… in the end, their usefulness when asked to do some bit of physics.

I teach the class from a certain text, at a certain level. I assign homework, which is where the real instruction takes place — as my students learn to do physics problems, apply principles, and untangle subtleties. I then design the test to be similar to the homework (sometimes the same problems!), but tweak the questions to test for comprehension instead of regurgitation. I provide a full formula sheet, in any case — they just need to know how to apply them.

As implemented, the students are never able to ace the test. Heck, few even finish. (I make this clear before the test and on the first day of class!). Why am I so cruel? I’m making a measurement, and I need to be sure the apparatus is sensitive along the whole band — low to high. Some excel — and I can *see* it! Others fail — and I can compare them to everyone in the class!

85-90% of the credit comes from longhand physics problems with no multiple-choice. It’s far, far harder to grade, and partial credit is crucial — but I get a real, individual picture of how the student is doing, concept by concept. And so do they. It’s also cheat-resistant — I notice patterns quickly when people try to copy, and finding someone correct to copy is challenging.

Then I curve the test. I center the normal distribution on 75%, then adjust from there (this only applies to large classes, for a large sample size.) If there are many A grades, that’s fine — I have no quota. I am measuring, and I’ll report what I measure.

In fact, the university at which I teach seems to have a pair of overlapping populations — the excellent students here for low tuition, and the run-of-the-mill population that follow the normal distribution. I *always* see a second peak in the A range… the subset that are dedicated and willing to do whatever is necessary to excel.

I allow the grades to exceed the point total on the test as well. If someone has truly excelled on a test, I let that carry over. The artificial cutoff imposed by disallowing this can bias the measurement.

I also compare the distribution directly to the homework distribution. If I see a large divergence between them, I adjust accordingly… but so far, I have never seen a serious mismatch.

If, someday, I have a class that is overloaded with world-beaters… each one can and will receive an A. The curve helps me normalize my measurement, but I am not its slave. I doubt that I’ll ever have to worry about seeing such a thing happen, though.

Finally, to address the question of “top students not studying together to retain an advantage”… what a crock. In any difficult topic, studying with other (often inferior) students is the *best* strategy, as teaching something to someone else will *cement* the concepts far better than just insular cramming. My very best students have been those who study together, and there’s never a concern about a limited number of A grades — those who earn them will get them.

I deem uncurved testing to be a carnival game (hit the bell, win a prize) instead of a measurement (what’s the velocity of the hammer head? The strike accuracy?). If anyone is hitting 100% uncurved, you’re either testing memorization (useful in some fields) or your measurement lacks dynamic range.

Leave a Comment

Please note that all comments will be reviewed before they are published.