The Weight of Grades
As we stumble our way into the future of technology in education, I’ve had the opportunity to meet some interesting people who have pushed me to think carefully about what I do. Last year it was Mark Gross, the CEO of School Loop, who caught my attention in a conversation about grading. While his web-based software supports weighted categories for grading, he’s personally against the idea. Now I’ve always used weighted categories: 50% of a student’s grade is based on tests, 30% on labs, and 20% on homework, (which is what I did last year in my AP class) or some other such breakdown depending on the subject and the grade level. Lots of teachers do this, and so it makes sense that School Loop would have the feature. So why not use it if it’s available?
Mark’s argument against weighting makes a lot of sense to me. He says that if 10 points on this assignment means something different than 10 points on that one, students get confused. And why wouldn’t they? We don’t set up weighted categories for the purpose of helping students clearly understand how they’re being graded; we set them up because it helps us as teachers control how we value various kinds of work. It gives us the flexibility to assign work and points here and there, knowing it will all be sorted out in the end.
But there’s another way: If I want 50% of the grade to come from tests, then I could simply be sure to plan that 50% of the points available in my class come from tests. If every point that a student could earn counted the same — test, quiz, lab, homework, whatever — wouldn’t that be more straightforward? Of course, this requires more planning and somewhat reduces flexibility.
In thinking about this trade-off, I realized I didn’t even really know what I had done last year. Were my homework points worth more than my quiz points? Or the other way around? I decided to look at what I gave first semester last year:
5 quizzes, with values from 10-25 points each, totaling 85 points.
1 test worth 100 points.
9 and a half labs, for 95 points altogether.
9 homeworks, worth 5 points each, totaling 45 points.
This adds up to 325 points. And if all points were equal, this would have put the grades at 57% test/quiz, 29% labs, and 14% homework. So because of my weighting, the points I gave for homework actually ended up counting for more than the points I gave on tests and quizzes. The breakdown wasn’t insanely far off from the eventual weighted values, but the final result also isn’t the whole story. There are also some strange situations that can occur along the way.
Let’s take “Johnny” as an example. He was typical of many students I had last year: willing to work hard on the programming assignments, but unaccustomed to the level of precision, deception, and outright trickery employed in my quizzes. So after the first month or so of class he had a C, based on high homework/classwork scores and low test scores. In fact, his homework/classwork scores were so high that getting 9 of 10 possible points on a lab brought his average for that part of the grade down. And when that weighted category went down, so did his total percentage. What kind of sense does it make to have a C in a class, score an A on a lab, and then have your overall grade fall as a result? It is of little consolation to tell the student his grade was previously artificially high because his best category wasn’t yet fully populated.
A system of total points, without categories or weighting, solves this problem. When a low test score drags the student’s average down, they can see it creep back up if they earn a better percentage of points on the next assignment. And all this requires is a bit of advance planning, along with some willingness to stick by that plan. Here’s what I’m going to do this time:
5 quizzes at 30 points each.
1 test for 100 points.
10 labs at 15 points each.
10 homeworks at 10 points each.
Add this all up and you’ll find that there are 500 points in the semester. This way, it’s instantly obvious that every 5 points represents a percent of the final grade. And the first thing that jumps out at me is the homework points. Last year I often awarded full credit for the homework if the student bothered to answer the questions at all. And maybe that’s fine, but now that I see that each assignment is worth fully 2% of the grade, maybe I’ll handle it differently. Another thing that jumps out is the distribution between the test and quizzes. Last year I put an awful lot of weight on that semester final, making that one test over 30% of the final grade. I remember needing to curve it and offer extra credit in order to pull the grades back into a reasonable general distribution. This time around it’s still a big deal, being 20% of the final grade, but that reduction feels a little better to me. This also helps me contextualize extra credit opportunities.
Overall, I am realizing that last year I didn’t really know how I wanted to assign grades. Many teachers say that giving letter grades on report cards is the worst part of the job. Personally, I think the worst part is waking up in time to make it to school before 8:00am. But no one, including me, relishes the requirement that we reduce everything a student has done and worked at and learned over a period of months into a single letter. Since it is, however, a requirement of the job, I’m looking forward to being more clear and direct about how that will happen. I hope that leaving categories and weights behind will let me and my students all be more comfortable with the reality that grades must be given. I also hope that, if communicated clearly, the points might even be a good motivational tool.