My last post talked about a way to have daily quizzes in my Standards-Based Grading (SBG) optics course. It (and the comments) got me thinking about how to do it even better and I think I’m closing in on a better plan.

The main idea is to have daily quizzes that are problems randomly selected from the previous day’s work. It reduces the amount of homework I have to grade, and tackles the cheating problem since it’s now a no-notes quiz. I liked it a lot in my fall class and I definitely want to keep those strengths. My suggestion was six problems per day that would act as the only contexts for any future assessments (quizzes, screencasts, oral exams, and office visits). One commenter noted that might be too much to ask the students to absorb from Tuesday to Thursday. Also, I wasn’t too happy about the double quiz I suggested on Tuesdays (one for the previous Thursday material and one to act as a re-assessment of week-old information). So, here’s my new thinking:

- Assign 3 problems per night
- Have them be substantial, covering various aspects of what we talk about in class.

- Each day do a quiz on a randomly selected problem from the previous 6 problems (three each from the last two days of new material).
- Have the students maintain a portfolio of all the problems so that they can act as context for all future assessments

Things I like about this:

- Finding 3 solid problems sounds much more fruitful (and easy for me) than finding six every day.
- I really like the portfolio idea. Want to come improve your standard score? Bring in your portfolio and I’ll randomly ask about one of those three problems. For each of the standards the students will (hopefully) be encouraged to really comprehend the issues around the three problems, especially given that they and I will be encouraged to “turn them inside out” for every assessment.
- Before every quiz they should be touching up six problems in their portfolio. Admittedly if the quiz is on one they’re not ready for, they get a crappy grade but they can redo it via screencast, office visit, . . .
- Something we’ll go over today might show up next time or the time after that, allowing for some cycling (we will likely discuss the context of the quiz beforehand and often the details of the quiz afterwards, especially if it seems people are unsure how to approach the problem).
- Three problems times ~25 standards is a workable number of problems that the students need to master (especially considering that they are in groups of three with common ideas). Certainly it’s easier than six times 25.

Things I’m not sure about:

- The students “only” have to know how to do three problems per day. Master those, and they’re guaranteed an A. I get student evals sometimes that say I need to do some sort of high stakes exam to make sure they really know it. I’ve tended not to heed such advice, but this has me thinking about that again.
- There’s a chance that a standard might not ever be quizzed (25% chance, I guess). That means that they’ll need to submit something on their own. I guess I could use my old “one week rule” (here’s a post back when I called it the two week rule) or something. I could also weight the random selections differently to reduce that 25% to, I don’t know, 10% or something.
- Hopefully the notion of keeping up a solid portfolio will lower the barrier to having them submit something.
- If I had the quiz be on the last 9 problems, there’s an even greater chance that a standard doesn’t ever get quizzed (29.6%)

- The days could devolve into “how do we do these three problems” instead of active learning around the content.
- Students might want to do their own problems for the oral exams (that’s how I’ve tended to do it) instead of just coming with their portfolio ready.
- A compromise could be that I’ll tell them which standard they’re going to be reassessed on and they can polish up those three problems, of which I’ll randomly select one to grill them on.
- Another approach could be “bring your whole portfolio to the oral exam and I’ll randomly select anything in there.” I think that would really drive home the notion of keeping up a good portfolio but they might rebel.

So that’s where I’m at (for today 🙂 Your thoughts? Here’s some starters for you:

- I think 3 is too many/few and that instead you should subtract/add x and here’s why…
- I’ve taught with a portfolio approach before and here’s where I think your system is going to fail . . . (this is a cue for my friend Bret to weigh in)
- You definitely should also have assessments that do completely different problems and here’s why . . .
- How would you teach the students to “turn a problem inside out?”
- Here’s how I’d solve the 25%-that-won’t-get-quizzed problem . . .
- I think for the oral exams you should limit what they’ll need to bone up on and here’s why . . .
- I think for the oral exams you should make everything on the table and here’s why…
- Why not have every quiz be a random selection from anything in the portfolio?
- Below is a histogram of running 1000 semesters and finding how many problems would never get quizzed using this approach. The average is just a little over the 25% that I get with my approach above

I tried something like this one quarter and for me it didn’t work. Maybe you’ll have more luck. The problem I encountered was that students had a very hard time talking through their portfolio problems. Usually they just wanted to read what they wrote, or at least rely very heavily on what they wrote. It was a much different experience than having them work new problems, or answer new questions.

Yeah, I can imagine that. I think the notion of requiring things to be “turned inside out” will hopefully get them to dig under the surface to really understand the concept. How many problems did they have in the portfolio by the end?

This was in a grad quantum course where the standards were broken up into three sets, each covering about 3 weeks of class. There was one portfolio review for each set, so maybe 12 problems (each with multiple parts) for each review. Maybe that was part of the problem.

very interesting. So maybe my more frequent assessment might help?

Three problems seems like too small of an amount to me, but that might be because I think there are way more than three problems that I would want _my_ students to know per standard. Your field is probably way different, though.

But I would be afraid that _my_ students would just get to know the three problems really well, and only know those (and most calculus problems I can think of aren’t rich enough to justify only three per standard).

I’m not as worried about that. Here’s an example: topic: polarization

1. Use jones matrices to calculate the output of a random light source through a 1/4 wave plate, polarizer, and half wave plate with arbitrary angles.

2. Explain how you can protect your laser from back reflections off of a laser-produced plasma assuming you can do the experiment with circularly polarized light.

3. Explain what happens as you watch a 3D movie by wearing the glasses wrong.

These are all pretty rich and interconnected. Certainly the day we do that stuff will likely just be many examples just like this. Your thoughts?

These seem like pretty good questions, and I think that I agree that three is probably sufficient. I just need to learn how to write better questions.

Pingback: F is for midterm | SuperFly Physics

Pingback: Physics majors practice presentations | SuperFly Physics