Wednesday, December 12, 2018

Curved grading

I have a beef about traditional grading. Let's use seminary as an example. Take comprehensive exams. Professors don't expect students to get 100%. Few students get a perfect score. 

Yet that means they don't expect students to remember everything they were taught. But what's the objective of teaching? The objective is not to pass the test, but to master the material. If, however, there's too much material to remember, then students aren't learning what you're teaching. But why teach more than they can learn? Why teach more than they can remember? Shouldn't mastery be the goal?

Consider an exam on internal anatomy, in which a med student scores 90%. Say that's a passing grade. 

Say that means he can identify 90% of your internal organs. It's the other 10% he's confused about. Would you undergo an operation by a surgeon who could identify 90% of your internal organs? What about the pesky 10%? Is that a kidney or the spleen? Flip a coin?

Someone might object that surgical knowledge is a life and death affair, so there's not the same margin for error. But does that mean seminaries teach students lots of irrelevant stuff? 

A better approach might be short, weekly quizzes. Passage requires a perfect score, but they can retake the quiz until they get 100%. At least that way they know it all. They mastered all the material. They remembered what they were taught. They may forget some of that at a later date, but that's better then never knowing it in the first place. 

Then there's the tradition of grading on a curve. That's unfair for two reasons.

To begin with, curved grading is, in effect, a group grade. You're not being graded on the quality of your own performance. Rather, your grade is dependent on how all the other students performed, even though you have no control over their performance. 

But it gets worse. Some students cheat. And that raises the bar for the honest students, since curved grading averages the grades. If students weren't graded on a curve, then each student's grade would be independent of the others. But if some students cheat, that raises the bar for the other students by distorting the curve. Honest students are downgraded by the cheaters. Honest students get a lower grade than they'd otherwise get because they're competing with students who take unfair advantage. 

I don't speak as a disgruntled student. I was a high-performing student in college and seminary. I'm speaking on behalf of other students. 


  1. Speaking as one who teaches professionally, what you are calling mastery largely isn't possible at higher education levels because the kids haven't been trained to do it.

    I only have them for a semester and that isn't enough time to undo a lifetime of education training that teaches superficiality. Plus that superficiality is still being taught them concurrent with my class because my class isn't the only one they're taking.

    Instead, I've found that I need to do the best I can with what I have. This has meant doing things like weekly quizzes instead of exams (somewhat like your suggestion), structuring the final so that it can help the student who has improved over time, etc.

  2. I was never good at memorization. Luckily, most of my engineering courses were open-book. And that reflects how real-life engineering generally is.

  3. 1. Every med school that I know about employs curved grading for many if not most their subjects. It's either that or subjective evaluations where one or two attending physicians determine your entire grade.

    2. In fact, the national licensing exams to be licensed as a physician are curved. Your score is relative to all the test takers who took the test at the time. And the score one receives on these exams strongly determines what specialties one can or cannot apply to. Hence these exams are super competitive.

    3. There's rampant cheating for many med school exams as well as as the licensing exams. Tons of material available online if you know where to look and tons of material passed around between med students. I'm not condoning any of this, but I am aware of it.

    4. Virtually every med student and junior physician knows the common things (e.g. common anatomy, common diseases, common drugs). What separates the students' grades or rankings is usually what's uncommon. There's a saying in medicine: if you hear hoofbeats, think horses, not zebras. However, when it comes to exams, it's more likely the zebras that are tested, since almost everyone knows the horses already.

  4. 1. A test in which large numbers of students get 100% surely means that you are only testing the lower elements in Bloom's taxonomy of learning: names, dates, identity of internal organs. But if you are to ask students to discuss the causes and results of the rise of Pietism, for example, how could you expect any Master's level student to give a perfect answer? A 300 page monograph would be unlikely to yield the perfect answer. I remember my own initial astonishment when I would get a perfect score on an exam question in seminary, having been raised in a British context, where 50% represented a pass and 70% was verging on excellent.How could thirty minutes of scribbling by a first year student represent the complete answer to any question on the Doctrine of God?

    2. In the light of the above, exam grades are typically relative rather than absolute. British exam grades do not mean that their students understand less than US students, but that the baseline against which they are measured is different. 50% does not mean "He understands 50% of the material" but rather, "he is about average among similar students". If every student has to get 100% to pass, how do you distinguish the more able student from the less able? And if a seminary student isn't able to give an exceptional answer about the rise and results of pietism, should he not be a pastor? Perhaps he has other gifts that qualify him as a faithful shepherd of the sheep?

    3. Of course, an MDiv doesn't qualify you to be a pastor: that's why we have Presbytery exams, which function like medical boards on a pass-fail basis: does this person know enough to be qualified. I don't know any presbytery that functions on a 100% pass mark.

    4. If 2. above is correct, then there is always an inherent baseline against which students are being measured in every test that is more than objective. And not all tests are equal, even objective ones. Sometimes I ask bad questions that disadvantage the students compared to last year's batch. Or perhaps I didn't teach part of the subject very well this year. If the grade is not an absolute but a relative indicator, is a student not a good student because of my failings? So I'll often grade a few of the better students' answers first to get a feel for what a "good" answer looks like. Of course, sometimes your expectations are upended and a good student bombs a question while a less able student nails it. But I have some idea of what "good" looks like. Of course, after twenty years of teaching, I can also compare with previous classes.

    5. quizzes and tests have quite different pedagogical purposes, as well as having the goal of assigning a grade. Quizzes keep the students recalling material frequently, which helps the learning process. Exams tend to test the ability to synthesize more. In languages, quizzes are far more important than exams, in my view. We're experimenting with computerized testing that enables students to take and retake quizzes until they reach mastery (which may or may not necessarily be 100%, depending on topic). There is always a limited amount of learning time, however and you have to decide what your goal is. I could design my Prophets class in such a way that every student attained 100% on the final. But only if I redefine learning as filling a bucket rather than lighting a fire: I don't believe the students would be better equipped for ministry as a result, and the brightest students would particularly suffer.

    1. Iain,

      I'm sure you were a stellar student, so you could thrive in a highly competitive academic environment. And it makes sense for seminary profs. and college profs. to be the intellectual cream.

      My concern is for students of average intelligence who don't have the extra margin provided by superior intelligence to overcome the curved grading bias–not to mention the drag factor of cheaters.

      Curved grading isn't a liability for intellectually gifted students because they have an IQ advantage that puts them over the top and compensates for any cheaters. They can always make the cut if they apply themselves. But average students don't have that margin to spare. The danger is to set up honest students for failure by creating additional handicaps.

    2. On an ancillary note, I salute how you and Peter Lillback rebuilt the OT dept. at WTS. That's long overdue.