16

One of the biggest challenges I face year after year in teaching—and one that seems to affect my teaching evaluations—is that students are convinced that they are doing poorly in my course, no matter how much I reassure them that this isn’t the case.

Part of the problem is that the university provides a standard grading scale, and many of the instructors in my department strictly implement, a grading scale where you need to earn 90% of the available points for an A, 80% for a B, and so on. A fixed, unalterable predetermined scale irks me significantly for a number of reasons:

  • Chasing after an arbitrary goal is a bad way to actually learn material. I’d rather students worry about being able to apply what I teach them later than the grade they’re going to get.
  • Difficulty must be pre-judged, and gauged accurately, to ensure a fair grade distribution. This means if I screw up the difficulty of a question, I end up “hosing” my students.
  • There is little opportunity to recover from one or two mistakes on an exam during the semester, particularly if they are worth a large portion of the grade each. I don’t think one bad day should screw things up for my students.
  • I could impose an alternate scale but then I’d be committed to that—same basic result just different checkpoints.
  • Because of the nature of the class, it’s easy to make a mistake early on in a problem that means you can’t solve the rest of the problem, and the result must unfortunately be a low score (a student might make an error early and can only get 5 out of 20 points, but a mistake later on might result in a grade of 18 out of 20).

So I use the basic scale, but adjust it in the students’ favor at the end of the semester to adjust for what I think makes a fair and equitable grading distribution, and most students do reasonably well, because that’s an accurate reflection of their performance overall. But no matter how much I tell the students in the class that this will take place, no one really seems to believe it. (A typical distribution will typically be 20% A’s, with most of the rest B’s and C’s.)

Is there a way I can convince my students that even though my grading is strict but fair, even when students don’t get very good scores on individual graded exercises?


There has been some issue about the amount of adjustment going on, so I should probably clarify. Typically, the curve amounts to about two-thirds to a full letter grade: so the borderline for an A would be in the low 80's rather than 90. This curving amount leads to a distribution that is about 20% to 30% of the students receive an A, with about two-thirds B's and C's, and a smattering of D's and F's.

There was also the issue of how much grading is done, and it caused some confusion because I worked in two different positions where the grading is completely different. For the course I'm describing in this question, the students receive grades on every homework assignment, take a series of eight quizzes, and have a project in addition to a final. So students know what the "worst-case scenario" will be if there is no rescaling, so there is plenty of opportunity for feedback and course correction. My issue is absolutely not with grading—it's with grading on a predetermined scale.

  • Comments are not for extended discussion; this conversation has been moved to chat. – aeismail Jul 6 '18 at 16:08
10

Dealing with the university "standard" is much easier than it seems, since they don't (can't) determine what a "point" is. So 90% of the available points is meaningless.

There is a form of grading called "cumulative grading" (maybe some other terms apply). Decide that the course as a whole is "worth" 1000 points. Create student exercises/exams/whatever that total to 1000. For example, Project A is "worth" 300 points toward the goal. When a student does an exercise they get a certain number of points up to 300. (See below for refinements.) Every piece of graded work has a value, possibly all different, and they add up to your total.

The list of tasks and their individual point "values" are listed in the syllabus. Students get to see the big picture, up front.

You can penalize for late work as you like.

The students always know exactly where they are. When they earn 700 points they know that they have a C or better - guaranteed. The University is satisfied and the student knows absolutely.

An additional element/benefit of this is that some students have many things to do and my class may not be the most important thing. They might be satisfied with a B if it gave them time to work on other, higher priority, things and they could do so without risk.

Important Refinements.

I always let students repeat/refine their work for regrading. They couldn't earn full points on regrading, but, say, up to 90% of the points lost at first grading could be returned to them if they improved the work sufficiently. This means, essentially, that if they got 270 points on that project, missing out on 30, that they could earn back 27 points leaving them only 3 points short of full marks.

Re-grading gave two advantages. One is that I never got whiny complaints about points. Second, when I thought a student (or team) really needed to revisit their work, I could give a relatively low score on the initial submission. They thus had more incentive to improve it.

You may not want to do the following, depending on your load and your students, but I didn't limit the number of times a student could re-do a piece of work. I made exceptions only if it got excessive as the down side, for a few students, is that they obsess over some early work, falling behind on the new. So it takes balance.

In computing summary grades, no student ever misses a mark by an insignificant amount. If someone is ten points out of a thousand short at the end of the term their grade is "rounded" up. This is perfectly justifiable as there is likely some (hopefully small) subjective element to any grading of projects and the like. I gave up on "objective" exams a long time ago, since they measure only poorly and can be greatly affected by other things such as student stress, etc.

Finally, at the end of the course, I looked at the grades overall according to the above measures. I asked myself whether this seemed to be a fair measure of what the group as a whole learned. Usually it was fine, but once in a while it turned out that the grading itself was somehow off, and the students were better than the curve suggested. I'd make an adjustment. No one ever questioned this, nor would they have any reason to. Tenure is a great thing, however, when you want/need to do something that the University frowns upon. At the end of the day (term) you need to be reasonable.

Note also that I had a reputation around the University of being very hard/strict/demanding. But the students always had a reason to feel good about themselves and how they did. This is, in some ways, nearly as important at the technical (Computer Science) things I taught them. Actually being more gentle than your reputation is a plus.

Also, note that the philosophy behind this whole scheme is "You aren't here to prove to me that you don't need to be here."

  • Seems almost like you and I have the same "method" in different robes. – Gypsy Spellweaver Jul 6 '18 at 1:35
  • I think there’s a bit of confusion from the comments, in part because I’ve worked in two different systems. The first system was the crazy German system. I’m now in the US. My grading system breaks everything down into its components and explains how each section is computed. They also get the official scale, plus a caution that I’m allowed to change it. (The issue in part is that nobody else really uses that discretionary power. – aeismail Jul 6 '18 at 8:13
  • 1
    @Buffy Objective measuring is at the discretion of the instructor, to a point. I don't know how it is achieved in non technical fields, but I'd expect the instructor to set some overarching criteria. In similar subjects in school (like arts class) they simply didn't test for anything subjective and made you write about art history or recite from memory for the grades. – Magisch Jul 6 '18 at 11:14
  • 1
    @Buffy: "What is 94%+ on, say, creativity?" - is this actually an issue with Magisch's approach, or a difficulty with judging creativity in general? – O. R. Mapper Jul 6 '18 at 14:43
  • 1
    @O.R.Mapper, my objection is to all grading schemes that seem "objective" and "precise" but which are not. Moreover, those that are completely objective measure things of less actual value than those other things that require experience and judgement to measure. So, creativity was just an example of something hard to measure. – Buffy Jul 6 '18 at 16:42
8

I found that being as open and transparent about the grading policy is important. In the past course I taught, students performance improved on testing measures as semester started. The midterm was most students' weakest grade with their final being much higher. I was up front about this on day one and told the students not to worry as the midterm would always be curved to where the average grade was an 85/100 (not counting the bonus question). I learned the benefit of this the hard way after my first semester.

Other tricks that I have learned that make grading less of a pain for students (with minimal effort on my part):

  1. Never put gotcha questions on exams. I tell students that there will never be a question on the exam with content not covered in the lecture/slides. Further I provide a review sheet with topics and an example of what is a good response to a short essay question and what is a poor response. What is on that list is on the test. If a topic is not on the review list, its not on the test.

  2. I am up front about curves and where students have traditionally done poorly in the course. There is never any surprises in my course. Surprises are for parties and presents, not university courses.

  3. I tell students that I have a record of student performance on different test items. If students from a class do poorly on an item that historically students have done well on, I curve that item. For example, if the test average is a 90/100 (above my test curve) but an item on the test had a 2/5 average when the historical average is 4/5, I will curve that item up. And everyone gets the two points. I tell the students that, in that case, the fault is likely mine in that I did not teach that topic particularly well.

In particular, I have found that point 3 is the most popular with students but also the one that comes up the least in grading.

In short, I have found that fairness in grading is largely about student perception rather than any mathematical fairness in the grading.

  • WRT point 3, note that there is a statistical measure (whose name escapes me) that measures for each question how students do compared to the overall score. In particular, if a question is marked incorrectly by high performing test takers overall, the question is probably invalid. Maybe a statistician out there can supply the name. It was a standard feature of machine graded tests. – Buffy Jul 5 '18 at 19:11
  • I believe you are discussing item response theory. I think its item discrimination but could also be very wrong as IRT is one of my weaker statistical areas. – JWH2006 Jul 5 '18 at 19:15
  • Could you explain what "85" and "90" mean? – Udank Jul 5 '18 at 19:58
  • @Udank 85 or 90 out of 500 no sorry 100 most likely.... – Solar Mike Jul 5 '18 at 20:03
  • +1 For the last sentence! I would go as far and say that fairness is even a feeling. – Dirk Jul 7 '18 at 6:27
2

Something I've occasionally done after grading tests and preparing a (handwritten) histogram of the grades (5 point increments was usually enough for me), and the grades were a bit low and I also felt the test was a bit hard, was to decide where A, B, C, etc. cutoffs would be and then scale (or add points, or something; just make it explicit and tell the students) the grades accordingly. In your case, just make the adjustment so that the A's are 90 and above, the B's are 80 to 89, etc. I don't recall what I called this (I haven't taught in 13 years), maybe nothing and I just wrote the raw grade down and put the adjusted grade next to it and circled the adjusted grade, but I suppose you could call this an "administrative grade adjustment". Since this gets written on their tests and recorded in your grade book BEFORE the tests are handed back, I would think you're not violating the guidelines you're trying to get around.

Incidentally, the reason I did this when grades were a bit low is that I didn't want, for example, 5 points on test #2 that covers hard stuff being counted equally with 5 points on test #3 that covers easy stuff when computing averages at the end, as this would be unfair to someone who put a lot of effort into preparing for test #2 on hard stuff as compared to someone else who put the same effort into preparing for test #3 on easy stuff, where the former gets less points than the latter for the same amount of work.

  • I wrote the sentence about the grading scale poorly. The university lists the scale, but my colleagues implement it strictly. – aeismail Jul 6 '18 at 3:41
2

Perhaps one option would be to supply your students with your past grading data showing how much you scaled previous classes. This would need to be anonymised of course, and you might even want to add some controlled contamination to the data to make it impossible for students to identify a past student, but it should still be possible to show them the distribution of the raw marks, and how the distribution of the scaled marks after your adjustment.

If this kind of data were supplied, it would give students an empirical basis to estimate their scaled mark from their present raw marks, or at least get a rough sense of what to expect. The upside of this is that they will have a more realistic sense of how they are doing, perhaps mitigating the problem you are having now. Of course, a corresponding danger is that you might decide to scale less than before in one year, and students might be upset, having expected a higher scaled grade. Still, if your problem is that students don't believe you, perhaps showing them the hard data is best.

  • 1
    Everything moves together, so I could list the shifts by semester to give an estimate of what’s reasonable without compromising any individual grade. – aeismail Jul 6 '18 at 2:50
1

To satisfy the students' need to know their current standing in the course and meet the university's requirement for a strict grading scale you need to create a different system. The design of the system needs to be based only on itself, not past methods, and a complete from the ground up redesign.

Caveat: This system does not work well in a strictly fact-based subject area, such as History. It is best utilized in an "applied" field.

The first point in the new system is to test the apprehension of the subject matter as opposed to testing their ability to execute a procedure. An electronic device can do the latter while the former represents what should be the goal of education. Many people experienced elementary mathematics courses where they were instructed to "show your work." Much as students dislike doing so, it allows the instructor to grade the work, not the final answer. It's possible to drop half a point for failing to "carry" the one a single time when grading the "work" while only being possible to add or drop the full 5 points when grading the final answer alone. Meeting this objective requires that you design your examination questions such that the work can be shown, and will help demonstrate the grasp, or lack thereof, the student has on the principle involved. You also must explain to the students that their work, which demonstrates their understanding, is the target of your evaluations. A wrong final answer does mean dropped points, as wrong is still wrong. Why, and how, the answer is wrong determines the amount of lost points.

With the redesigned exams, and other assignments, developed, the system needs to have "weights" applied to various types of work. Exactly how those are distributed will depend on how the course, as a whole, is designed, and how you have decided to partition the assessments. By way of an example, you could use the following:

  • Weekly quizzes (14), scored 0-10 pts: 1% of total (each) [14% in aggregate]
  • Lab assignments (4), scored 0-65 pts: 6.5% of total (each) [26% in aggregate]
  • Mid-term exam, scored 0-300 pts: 30% of total
  • Final exam, scored 0-300 pts: 30% of total

  • Final total is up to 1000 pts. Letter grades assigned per institutional/departmental standard percentage scales.

If the course syllabus includes your weighting, or break-down, of the final grade, the students will always know where their work-to-date stands in relation to the final grade. If your institution requires a grade of "C or better" for prerequisite courses, the weight of the final exam can be set such that earning such a grade without the final exam is a statistical anomaly. In the example given, a student would have to earn every point available for all work, quizzes, and exams during the term to obtain a 70%, or "C" grade without the final exam. Of course, if rework is allowed, then such a perfect score becomes less unlikely, and the weights would need adjusting.

The final piece of the revamped system is in how the student work is graded, and points assessed. As covered in the beginning, evaluation of the work, rather than the result, is a key element. Other elements are:

  • Not assigning a point value to each question
  • Evaluating the work from all of the students and marking errors without assigning points in an initial round (A "raw" score, not recorded on the paper might be helpful without being a requirement)
  • Using the aggregate of the assignment, establish the cutoff for pass/fail and making a curve to cover the remainder
  • Assign the points-earned to each paper based on the curve created for that assessment and the student's marks on it (Fractional points are acceptable, or assign a percentage rather than points)
  • The assigned points, or percentage, not the raw score, are the scores entered into the course record for the student

If you allow for the reworking of assignments, then the curve will have to accommodate the probability that most students will avail themselves of that option whenever the initial score is below some threshold. The threshold will be personal to each student, of course, while the average grade for the student body is likely to indicate a typical value. Universities with a highly-motivated student body might expect the threshold to be near, or above, 90%, while a nursing college might expect students in computer classes to have a threshold of near, or below, 70%.

The system will meet the institutional guidelines of a strict grading scale and provide transparency in the grading system for students and faculty. As long as it is made clear to the students that their "work" is the basis of your assessment, they will have the knowledge that your grading is fair, even if your evaluation of their work is strict. Always assigning the score to the graded papers, along with the syllabus-listed break-down, allows any student to see their individual progress toward their target grade (hopefully always an "A"). By grading their work you can see where they might be faltering. If the graded papers include your notes, or comments, on their work, they can also see what they don't grasp when they thought they did, and can either restudy that area, or ask you for help with it. Lastly, evaluating the work allows you to avoid the situation where a simple error in the beginning of a problem causes a disproportionately low score. You can also evaluate the students' knowledge rather than their test-taking skills.

  • I already implement much of what you’ve listed. Like I said, I know where the students make the errors that cost them and do my best to reward them for what they demonstrate rather than penalize what they get wrong. But there’s not much I can do if they go in one direction and the problem actually needs them to go in a different one from the very beginning. – aeismail Jul 6 '18 at 2:55
1

At others have alluded to, you will have trouble assuring students that they will receive a fair grade because the grading scheme you've suggested isn't actually fair. Literally no one knows even what the grades will be until the end when the adjustments are made.

Why do I assert that the grading is unfair? You have missed one of the purposes of grades, which is to help students determine where they need to expend their efforts during a grading term. Even if the final grades end up being a reasonable reflection of the final understandings of your students, the ones who do poorly did not really receive a fair shake, because they were missing a primary indicator that they needed to pay much more attention. The same can actually be said of the students who did well; without clear performance feedback from you, they have little sense of whether they are rising to expectations, and even if you informally assure them that they are headed towards a reasonable grade, that means very different things to different people.

(As a small practical aside, why are you only curving at the end of the term? At the very least, why not curve each assessment as you go?)

While it is true that a question can accidentally be too hard, if the grades are as low as your question seems to indicate, then it sounds like your assessments are systematically too difficult, which means you're probably not thinking hard about what you realistically expect from your students. Remember, all that curving at the end does is to prevent you from having to design real, reasonable, and achievable course expectations. You can keep the work at an unreasonably high level, students can consistently miss the mark, and you can kind of pretend that they're getting it. But this isn't good pedagogy. Something is missing here.

I might sound harsh, but I am not unsympathetic to you. I appreciate the instinct to push grades aside and focus on the topic itself, which is the thing that can change our students into better people, and is ultimately the goal of the instruction, but ignoring grades in this way creates needless anxiety for students who depend on those grades for scholarships, good relations with their families, graduate admissions, or class rank. I would encourage you not to hold your nose and think about grades as some sort of necessary evil. Instead, you can use them as an opportunity to communicate whether your expectations are being met. Grades are an extremely powerful tool that we wield to drive real education, perhaps the most powerful leverage that we ultimately have. Don't let this tool go to waste.

  • Apparently I've not explained things well enough. The students have the official scale as a baseline and I only adjust their grades upwards in the students' favors. The only thing I don't know in advance is how much the scale is moved up. This is not inherently unfair. I don't curve individual grades because it's just as arbitrary. Since I'd have to adjust for the final exam, I don't see how there's any additional benefit to scaling everything. (A B- usually becomes a B+, so the curve is not as harsh as what it seems like.) – aeismail Jul 7 '18 at 5:24
  • Also, my issue is not with grading per se. I use it constructively—I go over material when I find grades on a problem are lower than I thought, I try to analyze where students are grasping material. I bristle against the idea of doing what my colleagues do: just use the official scale without alteration. – aeismail Jul 7 '18 at 5:35
  • @aeismail It feels like you've not read my answer. Adjusting each grade my be as intrinsically arbitrary as adjusting at the end, but it gives more meaningful feedback about standing as the term progresses. Arbitrary-ness was not the source of the unfairness I indentified; it was timeliness of feedback and knowledge of outcome. If I need a grade in the B range, will a C+ accomplish this for me or not? How do I compare my need to study for your class with my need to study for my other class when I have a final exam in both tomorrow, but I don't know what grades I am coming in with? – Ben I. Jul 7 '18 at 20:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.