[LON-CAPA-admin] new randomizations for re-opened sets

Gerd Kortemeyer korte at lite.msu.edu
Sun Oct 16 13:28:20 EDT 2005


Hi Peter,

On Oct 15, 2005, at 6:05 PM, Peter Kovac wrote:

> I have a number of requests from faculty new to
> LON-CAPA who would like to re-open sets for students with one  
> excuse or
> another.  However, the answers have been displayed and we worry  
> somewhat
> that a conniving student or two might copy answers.  Am I right in
> assuming that a re-opened problem set does *not* produce a new
> randomization?

That is correct. It does not.


> The plan right now is to create a duplicate set and open
> it near the end of the semester for students with legitimate  
> complaints.

Hmmm ... frankly, I would not do that.

a) making a duplicate is messy and makes all the grading harder
b) you really want them to do the homework asap, so they can learn  
from it for the exams. What's the point of formative assessment at  
the end of the semester?

> Would it be worth submitting a feature request for a re-randomize PARM
> flag?


Submitting a feature request is never bad, but making this change is  
quite serious. A quick fix would not be likely, since so much hinges  
on correct randomization.


> I.e., is anyone else interested and is it a reasonably easy code
> change?

The change is pretty serious, but not necessarily big. The biggest  
problem is the potential of making a huge mess of it.

One thing one could consider is to have a numerical problem PARaMeter  
that gets added to the random seed of the problem. If faculty sets a  
different due date, they can also set this to some non-zero value,  
like 17 or 42. Since this gets added to the random seed, it changes  
the randomization. Setting that problem PARM back to zero would  
restore the original randomization.

Now, that parameter would need to be stored alongside with the  
student submissions, so one can reconstruct what they had on the  
screen at the time, it would need to be part of the feedback, so one  
can see which randomization of the problem it referred to, it needs  
to be worked into the grading interface, so one can do intelligent re- 
grades ... aahhhhhh ..... $@#*$@&!

Having written this, there is probably a more elegant way to do this,  
and you really want to get this right.

Also, I am sure some faculty user will get him-/herself into some big  
old mess with this new feature ...

What I do with my course:

* the answer date is 12 hours after the due date. Within those twelve  
hours, the students cannot see the problem at all, and I can freely  
open it back up. So there is an automatic "grace period" within which  
students can send me emails, etc.

* beyond that ... well, give the students a hard time that they did  
not ask you for an extension in a timely manner, and just do it. I  
mean, the homework is there for them to learn the stuff. If they  
decide that they first copy down the answers, then send you an email,  
and then type the answers back in ... so be it. Homework is not worth  
much in terms of points, and I think the students will know  
themselves that this kind of stunt will hurt them on the exams.

- Gerd.




More information about the LON-CAPA-admin mailing list