[edit: throughout this post, I am talking about weighting assessments within a standard, not about weighting different standards]
Sarah asked if ActiveGrade will support weighting different assessments different amounts. A this-assessment-is-worth-twice-as-much-as-this-one kind of thing.
Huh.
Here’s my thing: I already feel pretty weird when I give an assessment and say, “I can tell from this that you understand about 80% of how to find the roots of a quadratic polynomial.” Right? We have to accept that the most we can say is, “on this problem you did about 80% of what the solution required.” We can’t know what our students know – all we get is evidence about what they know. Descartes & Plato, right? I know: deep.
So when I’m already admitting that my assessment is only a grave-rubbing of my students’ actual understanding, and that I can’t know both my student’s skill level and his velocity, I feel a little sheepish saying that one of my assessments is more accurate than another. When I give an assessment, I’ve got to be willing to stand behind it as a valid measurement. Calling assessment B twice as valid feels like calling assessment A half as valid.
On the other hand, shouldn’t final exams be worth more? Big projects? It’s not too hard to find ways that a carefully thought-out paper might be better evidence than a pop quiz, but maybe that just shows that we shouldn’t give pop quizzes. Of course, you can also show why a pop quiz might be better evidence than a paper.
So, I’m conflicted. Those of you who weight different assessments differently: how do you decide what the weights are? Those of you who don’t: what problems do you run in to?
gasstationwithoutpumps
October 22, 2010 at 9:40 pm
I tend to use subjective weights (looking at the series of grades and eyeballing a summary), but I do sometimes use deliberate weighting of grades to get final summaries.
1) early assignments are often “warmups” and worth less than later more stringent assignments.
2) short assignments are worth less than longer ones (generally the longer ones are measuring more skills, which is not very SBG, but I prefer more real-world tasks).
Riley Lark
October 22, 2010 at 9:48 pm
Focusing on point 2:
But within a certain topic, do you think a bigger assignment should be worth more than a smaller assignment?
Let’s take short assignment A and long assignment B, both measuring skill X. A turns out to be evidence that the student is great at X, and B turns out to be evidence that the student is only mediocre at X. What grade do you give? What if the skill levels are reversed?
PS: I think you can give a big, complex project and still be SBG – just give out the separate scores for the different aspects of the project. If a student aces a program that shows understanding of 4 different topics, you can just give the highest score in all four topics, right?
gasstationwithoutpumps
October 22, 2010 at 11:25 pm
I want sustained performance. If a student can’t do both A AND B, then they haven’t mastered skill X. Later, more substantial assignments have higher weight because they may indicate that the student has learned something from the earlier failure, and because they may be testing a higher level of mastery of the concept.
I do allow students to redo assignments they are not happy with (up to the last day of class), but the workload of the class is high enough that few can redo more than 10-20% of the assignments.
While a complex project *can* sometimes be divided up into different standards, often it cannot be. When a student turns in a computer program that is poorly documented and doesn’t quite work, I don’t really know whether or not he understands the theory behind the program or not. I simply can’t separate the inability to express the idea as a program from not comprehending the idea.
Almost all my assessments suffer from this problem: I’m interested in students being able to express fairly complicated ideas (like Markov chains and entropy measures) in cleanly written executable programs. Writing skills, programming skills, math skills, ability to read instructions, creativity, … all get mingled together.
Basically, at the grad school level, I’m really only interested in their ability to put things together in a coherent whole—they’re supposed to have already been assessed on most of the separable things over the past 16 years.
Riley Lark
October 22, 2010 at 9:52 pm
Focusing on point 1:
I think the subjective weights method is a great way to go. Have you thought about making these early assignments worth 0%?
A big enough topic for a different discussion, maybe.
gasstationwithoutpumps
October 22, 2010 at 11:30 pm
Yes, I’ve thought about dropping the weight of the first assignment to 0. But then I’d have to fail even more of the marginal students, as some have that warmup as their only real success.
Students don’t always learn from their mistakes. I’ve had 2 students take the course 3 times (with essentially the same assignments each time). One managed to *barely* pass on the third try, the other continued to be far below the pass level after three tries.
I certainly would not tell students that the weight of a specific assignment was 0, as many would then not do it. For a lot of the students, it comes as a real shock that they can’t program a very simple task quickly after 3 full courses in computer programming. The wake-up call of the first assignment is essential to getting them to put in sufficient time on the subsequent programs.
Chris
October 23, 2010 at 6:37 am
I’m being purpsefully ungenerous here, but isn’t grad school the appropriate level to be more hard-nosed about grading? One thing I’ve seen very dramatically in my time here in a Chilean high school is how being Good At School is a skill like any other: being able to sit still and focus through a very boring, restricted day. A lot of very smart kids are not Good At School.
But grad school is different, isn’t it? Don’t we require of grad students that they be Good At School? It seems to me that there are plenty of things about grad school more demeaning and important to fix than the grading system.
I dunno, I guess I don’t have a clear response, and maybe I’ve just dealt with too many educated-yet-thoroughly-incompetent people. It’s just…it seems like we should apply the toughest standards to people who are supposed to be future scholars. Am I making sense? Or completely off-base?
gasstationwithoutpumps
October 23, 2010 at 3:36 pm
I don’t think I’ve ever been accused of being a lenient grader before—more often I hear the opposite charge. I am perfectly willing to fail a student who does not meet standards, even if my class is the last one they need to graduate (one student in that situation has failed my class three times—she seems unable to write even the simplest programs required for the course—I’ve no idea how she passed the prerequisite classes).
The courses I teach are mostly graduate or senior undergraduate courses, so you would think that the students would be “Good At School”, but some are still quite incompetent, and I’m not willing to do a social pass (though it seems like some of them have had 16 years of that).
I had a couple of undergrads in my office earlier this week who had passed 3 programming classes (the required classes for computer science majors, not the wimpy ones for non-majors) and still could design a write a short program that I considered a warmup exercise for the course. I’m not looking forward to this weekend’s grading!
Some of my best students have not been traditional students, but re-entry students returning to grad school in their 40s and 50s.
I’ve been part of the discussion on SBG, not because I think that the grading system in my classes is the most important thing to fix, but because I think about and review different aspects of my teaching each year. After much thought I ended up deciding not to make any changes to my grading system this year, since the negatives of SBG outweighed the positives for my classes.
This year I’ve made some other changes: I’ve managed to get more class participation in my fall classes, and I’ve introduced a new component to the how-to-be-a-grad-student course (students presenting methods from Teach Like A Champion).
CalcDave
October 22, 2010 at 10:00 pm
I don’t think it’s that the assessment is any more or less important, but the skill itself. I need to test students on DeMoivre’s theorem in precal because the state standards say I do, but I don’t think that concept is as important as solving trig equations. A student who doesn’t get DeMoivre can still move on and get other concepts well, but without understanding trig, they’ll be lost for the rest of the semester.
So, I’d weight that skill more to show its importance.
Riley Lark
October 22, 2010 at 10:02 pm
Sorry, I need to be more clear: my question is about weighting the results of assessments _within_ a standard. There are interesting questions about weighting standards themselves differently, too, but I agree that it’s a different question.
erik
October 22, 2010 at 11:25 pm
I am struggling with this too. I am going with summative assessment this term, which is averaging the last 3 grades in each skill / content objective ( US History) Does this allow a student to just sit back and wait for the last 3 assignments?
Riley Lark
October 23, 2010 at 8:59 am
The usual answer is, “if the student can just sit back until the end of the term, there’s a different problem.”
Let’s say you’re going to give 5 assessments, and only the last three will end up “counting.” How do you talk about the first two assessments in a way that convinces students that they’re worthwhile?
Sarah
October 24, 2010 at 9:28 am
I inspired a post?! The funny thing is, I didn’t even think about my comment (“Re: reassessing – could the teacher specify reassessment limits? Or have the ability to weight the various assessments for a given standard?”) being interpreted in quite this way – I was thinking about weighting assessments within a standard as a way of making later assessments count more than earlier assessments, to reward improvement/maintenence of understanding over time (and help students avoid a situation where an early low grade masks their later improvement)…
But your interpretation has my wheels flying on this vein of thought too now… and wondering if maybe I do want to be able to weight some assessments differently! (Especially when thinking about a cumulative semester final – I’ve been struggling with how to incorporate the results of that into my grading scheme as it currently stands… luckily i still have some time to figure it out
Riley Lark
October 26, 2010 at 5:01 pm
The thing is that once you start weighing assessments, combination methods like the decaying average, the most recent, and the maximum stop making sense. You can’t simultaneously say that assignment A is worth twice as much as assignment B and that more recent grades are more important.
gasstationwithoutpumps
October 27, 2010 at 7:41 pm
Actually, you can have initial weights that differ and then scale them down with age, without any formal difficulties in the model or the computation. Good luck explaining them to students and parents, though.