Our Grading System is Broken | Hacker News Reader

Our Grading System is Broken

tshaddox.com

28 pointsbaddox37 comments

These are interesting suggestions. There is much more about the arbitrariness of grading systems in a book I read in high school, Wad-Ja-Get? The Grading Game in American Education.

http://www.amazon.com/Wad-Ja-Get-Grading-Game-American-Educa...

That book makes the excellent point that EVERYTHING about grades is arbitrary. Having a uniform system of accumulating scores on class work to set a course grade, as the author of the submitted article suggests, leaves a lot of important improvements undone. How does any outside observer know whether, for example, Podunk High School or Elite Prep Academy has a chemistry class that really covers the fundamental principles of secondary-education-level chemistry? How does an outside observer know whether an English teacher grades mostly on the basis of thoughtful argumentation and carefully chosen content, or on neatness and spelling only?

Any reform of school grading will be hard put to eliminate the role of standardized testing for precisely this reason. Harvard's dean of admission recently commented on this:

"Q: You recently lead a high-profile commission that recommended de-emphasizing the SAT and ACT from admission requirements. A number of colleges have already made the tests optional. Do you see this ever happening at Harvard?

"A: We do not foresee a time that Harvard would be test optional. Only a few years ago we were receiving applications from about 5,000 high schools each year and now the number has grown to over 8,000. We need some common yardsticks that enable us to gauge in a rough way what is being learned in an ever-increasing and diverse high school context, not to mention the increasing number of students who are home-schooled.

"We continue to believe that the College Board's Subject Tests, along with either the SAT or ACT with the writing tests, allow students the best opportunity to demonstrate what they have learned thus far. Advanced placement and international baccalaureate results are also helpful."

http://www.boston.com/news/education/higher/articles/2009/02...

Precisely! This is why I've always liked standardized tests (although admittedly, I did far better on my ACT than on my high school grades). Unfortunately, there are many arguably legitimate complaints about standardized tests, namely: that they're governed by the state or federal government, and that they may encourage schools to "teach to the test."

"Making the tenth of a percent from 89.4 to 89.5 worth immensely more than the 9.4 percent from 80 to 89.4 is idiotic."

I got an 89.4% in a CS class and asked the professor, whose office hours I frequented, if he could bump me up to an A since I was so close.

His response was: "If you truly understood the material and deserved an A, you would have earned an A by a large margin rather than missing it by a small one."

And what of the person who correctly guessed, or perhaps had a program misgraded slightly in his favor and made the A?

I agree with the sentiment, which is why I liked some of my professors who graded in statistical clusters (of course, economic professors). They graded on a curve, so everyone's grade was bumped up by the process, never down - and the clusters of A's, B's and C's were usually obvious. If you were in between a cluster (an outlier case) most gave you the benefit of the doubt and bumped you up.

Of course, I may have liked it less if I wasn't always in (or above) the top cluster =P

Then your teacher is admitting that my design only the few tenths of percents surrounding the letter grade cutoffs are worth anything. I suppose it's just a matter of opinion, but I find this system remarkably arbitrary.

Even if the grading system is arbitrary, it's usually fully transparent and disclosed ahead of time.

I knew that the cutoff for an A was 90%, and I also knew that the grade would be calculated based on my projects and exams. I knew the exact weight of each project and exam too, all within the first few days of class.

I knew what was expected of me in this situation and didn't deliver, so it was my fault for not getting the A - not the grading system's fault. But my 89.4% was so close to an A that I somehow felt entitled to getting one, but as his quote points out, I didn't really deserve it.

For college, which is voluntary, full disclosure would excuse the practice, although I still think it's foolish. However, in high school, which isn't voluntary, and where grades can affect your ability to get into college, it's worse.

I think the point isn't whether or not it's ethical (full disclosure in college would make it so), but whether it's a smart way to do things. I claim it isn't.

Well colleges use a combination of factors--grades, class rank, standardized-test scores, extracurricular activities, football talent--to decide who they are going to admit, and the practical difference between a first-tier and second-tier college degree is much less than most people are willing to admit. So one flukey B is not going to doom you for life.

There's always a cutoff in a discrete system.

Higher percentages should count a lot more than lower percentages. It is more difficult to go from 94-98 than it is to go from 54 to 58.

It seems to me that learning material over a long period of time is better than just "crashing" for exams, which this author's proposed system would encourage.

It is easy to imagine that a student will skip an assignment at one point, then stop doing homework altogether. The student will realize he should have done the homework toward the end when he realizes his grade, but by then it will be too late -- his homework average will be too low to make a difference.

His suggestion just saves the people who already know the material or who learn in other ways than homework from being penalized by having to do a ton of pointless busy work. If a person who thinks they can learn the material without doing the homework ends up trying to cram and then failing the test, that's his/her own problem. That person should just do the homework unless they're sure that they can learn the material without the homework.

Exactly. Struggling students already have the option to not do their homework with the existing system, this new suggestion doesn't change that.

I don't think it is only their own problem. Many people in school simply cannot recognize when doing the homework is necessary to learn the material versus when they already sufficiently understand it. Good teachers actually teach students how to learn effectively, and can help recognize how a student learns best.

It seems that the prevalence of homework => test in schools today is a result mainly of larger class sizes requiring the teacher to make compromises for students with varying levels of ability, motivation, and parental involvement.

Then they should just always do the homework and be on the safe side. Anyway, even if they screw up once and think they can get by without the homework, that's one F and they'll quickly figure out that that approach doesn't work for them.

The point I was trying to raise is that there are people who don't care if they get an F, but who teachers still have a responsibility to teach.

There's no sense in prescribing a change to "our grading system" without even referring to the content of the class that's being graded, based on the assumption that all the work a teacher gives out to students can be classified as "homework", which is for the purpose of practice, and "test", which is for the purpose of comprehensively measuring one's performance.

For example: If the teacher assigns a 10-page research paper or critical essay, most people wouldn't classify the paper as a "test", so under this schema it must be "homework". But I don't see how the skills that one could demonstrate by writing such a paper could also be demonstrated in an exam.

Actually, I would consider the large research projects to be the tests of that course. Besides, the English/lit/composition courses I've taken don't really give small daily or regularly assigned "homework" assignments anyway, so this plan wouldn't really apply to those.

What if the teacher feels that ten homework assignments can cover more breadth than two or three tests, and therefore can give a more accurate picture of the student's abilities? Would you also redefine those homework assignments as "tests"?

It wouldn't work. Homework is there so that you learn the material well before the exams instead of cramming. If it doesn't count towards the final grade, people won't do it.

I had a professor who said that having frequent (like twice-weekly) quizzes, which has the some purpose as assigning homework, had reduced the number of failures in the class considerably.

But with this system, the "safe" option is still to do all the homework assignments. If you're implying that making them option "punishes" the "bad" students (by bad I simply mean more likely to fail), then by that logic requiring the homework punishes the "good" students.

It's not about punishing anybody, it's about getting them to learn.

I strongly believe that having a grading system where a number of people need to have A for the grading system to be considered fair is harmful. In the university I went to in France, exams are made so that the gaussian curve of results in centered around 60% (12/20 in the french way of counting) with grades under 40% considered to be failing grades. Once you start having this kind of grading system, it forces the examinator to make the test hard and for that the best way is to have long tests (around 3-4 hours) where you have to apply the knowledge you gained on small projects (example a compilation exam where you have to work on a simplified subset of a natural language). Added to this, there was a healthy distrust of multiple choice questions and teachers who asked too much of them were considered by everyone as being lazy.

Contrast this with tests I took in the US at RIT where the questions were "When was the OpenGL committee founded?" (for a computer graphics class) or other such useless questions.

Homework was also different between both universities... In France, the graded homework were mostly medium to large size projects that took quite a few weeks to complete. In the Us, there was a much a higher number of busy work with small simple to do exercise that didn't really make student learn a lot (although I did have a few interesting homework projects so not everything was bad)...

Of course, the assumption here is that grades should represent how well a student knows the material, not how hard they worked in the class. I know a number of teachers I had used that philosophy, despite the clamoring of the kids at the top of the class.

Now that I'm not there anymore, I actually agree with them. Classes don't exist simply to teach course material - there is a lot of stuff you learn simply by doing all the work. Consider a lit course. The stated goal of the course might be reading and analysis of the course texts, but you also get better at writing simply because you did the "busywork".

The same holds for quantitative classes. There's only so much you can test in an hour, even if there are tests every week. Longer problem sets cater to a different skill set that tests, and they are both worth learning. A class that lets you skate by on pure test-taking ability misses a large chunk of the education.

I wish more classes used "portfolio assessment" to rate their students...

http://www.unm.edu/~devalenz/handouts/portfolio.html

...but given how labor-intensive it is to grade a portfolio, I can't blame teachers for shying away from the technique.

I was a full-time student for about 20 years, all said. I was never assigned a homework assignment that did not benefit me in some way. Whether by teaching me something new, honing a skill, or reinforcing and deepening knowledge I already had, each homework assignment I did was helpful. It seems like the author wants to cut out all of this beneficial work just because he can do well on tests. I don't think that's a good idea.

Your null hypothesis (or whatever it's called) is very pessimistic. Even if you benefited from those assignments, you might have benefited more from doing something else instead. Shouldn't it be your choice, as long as you can demonstrate command of the material?

In that case it would be a pre-testing strategy, not a post-testing strategy. If you have these pre-tests, you can then put more people in the right classes anyway, which would allow them to use their time well. you could then do something else which benifits you more, like a higher level, or faster class.

My assumption is that the teacher should optimize the class on the amount of the course material the students learn, and how well they retain that knowledge. As far as I can tell, the goal of all of the teachers I know is to get the students to learn the material the best they can, so they structure the class around that ideal. I fully understand that some students just want the credit, or to just get by with whatever they deem to be sufficient knowledge, but I don't think it would be generally beneficial to structure the class to give them incentives to do so.

In Sweden we have three grades: fail, pass, and pass with distinction. I rather like it as I don't believe it's meaningful to try to distinguish hundreds of different grades.

Engineers have one more grade than the rest of us: fail, 3, 4, 5

Making homework optional across the board (e.g. you can get an A if you ace the exams) is silly. It makes sense for a few courses (e.g. freshman writing, calculus sequence, intro language) which exist to train a specific mastery that can be objectively tested in a short period of time. It doesn't make sense, though, to pass a CS major based on exams alone (as opposed to programs), an English major without papers, or a math major without take-home proof assignments. He could argue that such assignments are tests, but then the distinction between tests and homework becomes so blurry as to be useless.

Making the tenth of a percent from 89.4 to 89.5 worth immensely more than the 9.4 percent from 80 to 89.4 is idiotic.

I'd argue that the problem is inherent in the fact that different grading scales (0-100 vs. A-F) are appropriate for numerically-graded exams and subjective assignments like papers. Converting a 0-100 scale to A/B/C/D/F has the problem the OP mentioned. But when it comes to papers, most people can only judge 4-5 levels of quality before inconsistencies pop up. This is one reason why the "E" (high fail) level of the original A-F grading system almost immediately dropped away, and D is rarely used. A/B/C/(D)/F exist, in effect, to mirror those separable levels of quality.

In fact, I'd argue that innate this 4-level limitation is one of the reasons many colleges (esp. Ivies) have such grade inflation in qualitative courses: with plus/minus grades, the three discernible levels of respectable work are mapped to A, A-, and B+/B, the one of mediocrity to B-/C+/C, depending on the professor's proclivities, and the truly awful work to C-/D/F... whereas a solid B (85%) in an intro science or language course is still a respectable grade.

I doubt the author of the blog post is saying projects and papers shouldn't be assigned. He's probably just saying that normal homework is a complete waste of time for some people.

I personally abhorred homework and loved projects and tests.

I agree with your statement about only being able to judge 4 or 5 quality levels. This of course only applies to subjective work, not to things like math, grammar, etc. Subjective classes like Composition usually don't feature such an emphasis on daily or otherwise regular homework assignments, therefore the large assignments would be the tests. But I would also argue that 4-5 levels of quality is too much, because is the grader comparing papers to other papers, or to some ideal paper that exists in his mind? I think for subjective grading, the only options should be pass or fail.

But I would also argue that 4-5 levels of quality is too much, because is the grader comparing papers to other papers, or to some ideal paper that exists in his mind?

I think, in general, the grades A, B, and C map to the following standards of acceptable work:

A: That of a student intending graduate study in the field. B: That at a major. C: That of an ordinary student, e.g. a non-major.

Hence, C is failure in graduate school, whereas a C in Calc 2 for an English major is unambiguously not failure.

F obviously represents failure. D is essentially a high-fail that's counted as a pass. It means, "you didn't pass this class, and you shouldn't use it as a prerequisite, but you didn't do so poorly that I want to set you back a course." D would probably vanish if college weren't so fucking expensive.

Some schools already do something like this. For instance, I know that Brown does not have pluses or minuses, and they only have grades of A, B, C, and NC (no credit). Furthermore, the NC grades do not appear on official transcripts. On top of that, you can take almost any course Pass/No-Credit with an option of a written evaluation instead of a grade.

Why does the 89.4 to 89.5% matters anyway? It will all average out over the 40 or so classes one takes in a typical B.Sc.

I've personally felt that grades for a class over the course of a semester (and also from freshman year to graduation) should be calculated using an average of both the student's overall grade point average as well as some sort of moving average, that shows a more realistic progression of their knowledge.

The whole point of a class is to go from a point of lesser knowledge to a point of more knowledge on a subject, but by simply averaging out student's scores over the course of the semester, there is no evidence of their progression. By using a combination of a moving average and the total average, a teacher can more accurately see how well the student is progressing.

Reply on news.ycombinator.com