How does grading work when we have multiple lab units under one subsection?
I have a subsection with three Lab units . At the subsection level I configured Grading with “Grade as: Lab”.
I have one Lab Assignment Type in Settings → Grading with a Weight of Total Grades set to 100.
I would expect that for each Lab unit where I pass the Lab, I get a 33-ish% score.
What happens is that when I pass a single Lab Unit, I get a Lab score of 100% and a Total score of 100%. This is without even attempting the labs in the other two units.
After attempting all three labs, I reset my score to 0, and try again to submit answers for all three Labs. At this second try, I get the behavior I expect:
after I pass the Lab in the first unit I get a 33% Lab and Total score
after I pass the Lab in the second unit I get a 67% Lab and Total score
after I pass the Lab in the third unit I get a 100% Lab and Total score.
If I chose to delete the learner’s state, I’m back to getting full score of 100% for the first lab without even attempting the other two.
I’d really appreciate some help here. I’ve been struggling to understand and define this problem so much that I dreamed about it last night.
I have setup a course with single Lab Assignment and it works as expected. My all 3 units are in single subsection. Here are my Lab assignment settings
Hi @elenalindq - Could you possibly share your grading setup from Studio and ensure that all is set how it should be (i.e. no droppable, correct number of assignments with the correct grade weighting)?
I expect you’ve probably done this, I just want to make sure the basics are covered before I start wildly speculating
Unfortunately I’ve finally had a chance to look at this and I’m thus far unable to replicate it on my end - I’m getting 33% as expected on my library content every time. I can only assume it has something to do with the interaction between deleting the learner’s state and the content libraries. Ugh. I’m afraid I’m at a bit of a loss, and would recommend speaking with someone who is able to review the state for those problems from the back-end to work out where exactly it thinks that 100% grade is coming from.
As far as I can tell from what you’ve said and provided, nothing about this looks wrong from the frontend perspective, so I can only assume there’s a bug somewhere that I’m unable to replicate. Notes from my attempt at replicating it in case they prove helpful at all:
I set up a timed exam with 3 units, and grading as shown above
In each unit I included some problems from 3 content libraries using the randomized content block
I registered a learner and enrolled them in my course
I completed one of the units and submitted my timed exam
I confirmed that on the progress page I earned 33%
I switched to my instructor account and reset my learner account’s timed exam attempt
I confirmed on my learner account that I now have 0% progress
I completed the timed exam again, this time completing everything except the content I completed the first time, and got 66% as expected
I can only assume it has something to do with this specific content, or something introduced in a version we’re not currently using (The site I tested on is still on Hawthorn).
Thank you so much for your detailed answer and for the time you spent checking the problem I reported. I deeply regret that I did not get back here with updates once we did fix this problem, which means you spent a considerable amount of time looking into something that I found an answer to. I’m really really sorry about that and I promise I’ll do better next time, and update my questions once I have progress. Lesson learned!
The problem I had was when using an XBlock and not a Problem type. I actually used the Problem type behavior to get a better understanding on why the grading is not working correctly when using the XBlock.
The XBlock did not implement the ScorableXBlockMixin. Implementing the methods defined by the mixin fixed it, and I get the correct grade calculated on the first try now.
XBlocks that don’t implement ScorableXBlockMixin will show the same behavior I wrote in the problem description in Juniper and (presumably) Koa.
