Wednesday, March 21, 2012

Making the grade


I have a couple of problems with how grading is traditionally done in education. One is the whole notion of a grade spread from, say, A to D. Shouldn’t the pass/fail cutoff come higher?

The implication of a wide grade spread is that teachers are teaching unnecessary material. Teaching more than students need to know. After all, if you can get a lot of questions wrong, but still pass, then, by implication, you don’t really need to know the stuff you got wrong.

Yet surely it’s important (depending on the course) for students to get some things right, and not just randomly. To take an extreme example, imagine a surgeon who could identify 5 out of 6 vital organs. Would you want him operating on you? Does he need to know all 6 vital organs? Likewise, there are several core problems that an automechanic needs to know how to diagnose and repair. What if an automechanic knew everything about a car except how to identify or fix faulty brakes?

Shouldn’t exams focus on mastering the essentials? If you can get a number of questions wrong, but still pass, or even get a high score, doesn’t that mean the exam is padded with unnecessary questions? Isn’t it better to keep taking the same test until you master the essentials?

I also have a beef with curve grading. If teachers didn’t grade on a curve, then even though cheaters would still have an unfair advantage, they wouldn’t push the grade down for honest students.

But when teachers grade on a curve, that averages the scores, so the grade distribution penalizes honest students. The grader totals the scores, then divides by the number of students. This means students aren’t graded on their independent performance. Rather, it’s a comparative grade. You’re graded relative to other students. 

Every time the same test is given, there’s a different grade distribution. You could get the same questions right or wrong, yet you’d receive a different grade each time you took the same test. Isn’t that arbitrary?

And the cheaters effectively downgrade the honest students, because their scores skew the grade cutoff (from A+ to A to A- to B+ to B to B-, &c.), since curve grading centers on a mean score. The high scorers depress the grade distribution for the lower scorers. When you throw cheaters into the mix, who score high through cheating, they win by making other students lose.

6 comments:

  1. Steve Said: Shouldn’t exams focus on mastering the essentials? If you can get a number of questions wrong, but still pass, or even get a high score, doesn’t that mean the exam is padded with unnecessary questions?

    This assumes that the answers to the questions are necessarily the point of the exam.

    Speaking for myself (an instructor of undergraduates), the quizzes, exams, and other assignments that my students are responsible have more than one end.

    Correctly answering the questions based on memorized material is one end. Being able to synthesize material on the fly to answer other questions is another end. Doing the best you can in a limited amount of time is yet another end.

    There are more, but you get the point.

    ReplyDelete
  2. MR. FOSI SAID:

    "Being able to synthesize material on the fly to answer other questions is another end."

    I think the take-home essay is better for synthesis.

    ReplyDelete
  3. I won't argue with an opinion. :^D

    ReplyDelete
  4. As a tolerant man I firmly believe that every issue has two sides: my side and the wrong side.

    ReplyDelete
  5. LOL!

    Having frequented this blog for some years now I would never had guessed that.

    ReplyDelete
  6. Largely, I agree. There are a number of factors involved that complicate the accuracy of grading, even objective grading.

    If you have more than a couple of students, you have a spread of aptitudes. For classes where subsequent classes require certain information be mastered in preparation, there is a minimal amount of information that objectively must be learned by the students. Classes are - or should be - designed for a high enough percentile of students to learn the minimal amount of information necessary for entry into the subsequent classes. But most students would do better having additional or tangential information for a more well-rounded understanding of the information beyond simple recall of it.

    The problem is that testing isn't particularly designed to cull students who don't get the minimal objective information if they can make it up in the tangential stuff.

    But most subjects have some level of subjective ability, understanding, or retention. History, for example, as part of a base education, requires only a general understanding of the flow of events with the ability to regurgitate some modicum of facts. Composition is fairly subjective holding to an objectively recognizable form, and typically requiring an objective usage of basic grammar. (Grammar in advanced composition is more subjective contingent on editorial decisions.)

    So grading on a curve may not be entirely beyond the pale except for your reasonable observation regarding cheaters. Grading on a curve is partially subjective. I hate that someone may make a worse grade for the same work if the class is loaded with exceptional students.

    But fully subjective grading should be shot dead, burned, and its ashes fed to hare-brained educators. My last year in Bible college was frustrated by an experiment with subjective grading. Not only was everything on a curve, but a certain spread was required. So if two students produced the same work and the spread demanded that one get one grade and the other get a certain lower grade, you had to make a decision who would get what grade outside of the work they actually did.

    ReplyDelete