[LON-CAPA-users] Bubblesheet Grading: Use same CODE for a copy of a course folder?
Stefan Bisitz
st.bisitz at ostfalia.de
Thu Feb 21 13:41:10 EST 2013
Thanks for your help so far.
I try to follow your suggestions - still fighting with some other major
issues in the background, so I'd like to apologize if I'm not capable to
fully follow all the details.
One important hint:
> The copied folder will contain different
> symbs, so the randomseed will be different for a given CODE, and the
> randomization could therefore be different [...]
So, we cannot use this way. Would have been nice. I understand why this
cannot work.
> [...]
> Note: to compute Exam 2 scores you will need to combine totals from both
> Folder A and Folder B. This suggested approach also assumes that a
> given student was only assigned to either Exam 1 or Exam 2.
Seems this is the problem. Some of them already took part at the first exam.
I try to explain what happend:
Exam 1:
Two weeks ago, Folder A, CODE 1111 and 2222, 120 students, randomly
assigend one of the two assigments (some with CODE 1111, some with CODE
2222), already graded. Fine. Done.
Exam 2:
- Plan:
Today, Folder B, CODE 3333 and 4444, 120 students (mostly the same as in
exam 1!), randomly assigend one of the two assigments (some with CODE
3333, some with CODE 4444). Later on: grade.
- What actually happend:
One random half got assigned the assigment with CODE 3333 (that's fine),
but the other half got assigend with the assigment from exam 1 with CODE
2222.
Please be aware of that some (statistically one forth) of the latter
students already did the exam 1 / folder A.
The best solution we could think of would be to grade the data for the
latter case as a new assigment which needs to be created now. This
appears not to be possible.
So, I guess Stuarts solution is not possible. Any chance?
"Half-Manual" grading, as Ray suggests, might be possible. We don't need
the students to see the grades online though having the data in LON-CAPA
would be good, e.g. for build-in statistics and so.
Thanks
Stefan Bisitz
Am 21.02.2013 18:00 schrieb Raymond Batchelor:
> I am also unsure about what *exactly* happened.
> What I do believe, is that the exam questions generated for any
particular CODE are also dependent upon the exact course and sequence
from which it was generated.
>
> So, it seems easier, to me, to move students rather than questions.
>
> Thus, my solution in a similar situation (since the students have
already answered the questions) might be to enroll these students in
both courses from which their exam questions were generated. One could
set these students as enrolled "in the past", so as to distinguish them
from current students and prevent them from seeing the "process".
> Unfortunately, it sounds to me like the switched questions were only
*some* of the questions from a particular sequence, rather than an
entire sequence.
>
> Of course if there are only 4 different exam versions, it might just
be simpler to grab the key answers for those versions and grade the
students outside LON-CAPA in a spreadsheet or database (which is what I
like to do anyway). Of course this will not satisfy you if you were
expecting to open up those questions for the students in LON-CAPA afterward.
>
> Perhaps someone who is intimate with the programming of this can
suggest how you might make one course-sequence mimic a different
course-sequence.
>
> Not me .... though I am intrigued as to what the best solution might be.
>
> We had something like this happen once, but it was an entire exam
taken from a previous course. We fortunately did not plan to open up
those questions to the students online, ever. So, we just enrolled the
students in the original course, graded them and output the scores from
Chart, for inclusion elsewhere.
>
> Good luck with this.
>
> Ray
>
>
Am 21.02.2013 17:51 schrieb Stuart Raeburn:
> Hi,
>
>> I am not sure about the way how the CODEs and randomization are
>> created.
>
> When a CODE has been specified, the random seed to be used is generated
> from a digest of the CODE, the courseID, and the identifier for the
> unique resource instance (known as the 'symb').
>
> Whenever bubblesheet grading is used to grade an assignment the CODE
> used during the grading pass is stored in the student's record for that
> assignment. Subsequently, whenever that resource is rendered for that
> student (e.g., student viewing answers etc.), the CODE will be retrieved
> and used to re-generate the random seed for that particular instance of
> the resource.
>
> If I understand correctly your description of the situation you are
> facing, a student taking exam 2 received a printed exam of (let's say)
> three pages, containing the following components:
>
> Page Origin of questions CODE used in printout (and shown in header line)
> 1 Folder B 1345
> 2 Folder A 2262
> 3 Folder B 1345
>
> where page 2 was a page that also occurred in exam 1, and the student
> bubbled in a single CODE for this exam -- 1345 -- i.e., the CODE shown
> on the first page of the exam.
>
>> My idea:
>> - Copy the course folder of the first exam
>
> I don't see this working. The copied folder will contain different
> symbs, so the randomseed will be different for a given CODE, and the
> randomization could therefore be different from that which applied when
> page 2 was originally printed for exam 1 (but mistakenly included as
> page 2 of exam 2).
>
> The solution I would adopt would be to:
>
> 1. Create two raw bubblesheet results files, by separating the student
> responses into those for included items from Folder A, and those from
> Folder B. StudentIDs, names etc. will be duplicated in the two raw data
> files.
>
> 2. Upload the first of the derived raw data files to the course via the
> bubblesheet grading menu.
>
> 3. Hide the resources in Folder B, which were missing from the printed
> exam 2 students received.
>
> 4. Grade the Folder B exam using the first derived bubblesheet file, now
> containing only bubbles for the resources from Folder B, with the
> checkbox checked for "Skip hidden resources when grading".
>
> 5. Replace the CODE in each line of the other derived bubblesheet raw
> data file (i.e., change it from 1345 to 2262)
>
> 6. Upload the second derived bubblesheet file, containing only bubbles
> for the resources from Folder A, via the bubblesheet grading menu.
>
> 7. Hide the resources in Folder A, which did not appear in the printed
> exam 2 students received.
>
> 8. Grade the Folder A exam using the second derived bubblesheet file,
> now containing only bubbles for the resources from Folder A, with the
> checkbox checked for "Skip hidden resources when grading".
>
> Note: to compute Exam 2 scores you will need to combine totals from both
> Folder A and Folder B. This suggested approach also assumes that a
> given student was only assigned to either Exam 1 or Exam 2.
>
> Stuart Raeburn
> LON-CAPA Academic Consortium
>
>
> Quoting Stefan Bisitz <st.bisitz at ostfalia.de>:
>
>> Hi Bubblesheet graders,
>>
>> Your help is needed :-)
>>
>> We have two similar CODEd Exams, each with two different 4 digit
>> numerical CODEs, all in one course, two folders. The PDF files have
>> been printed. Bubblesheets have been filled in by the students
>> corresponding to the questions in the PDF files. The sheets have been
>> scanned and then uploaded to and graded in LON-CAPA.
>>
>> We unfortunately had a mistake in the chain of process of one of our
>> Bubblesheet exams. It turned out that one part of the assignments for
>> the second exam have been switched with the questions (and CODE) of the
>> first exam! Hence, we stopped the validation check for the second exam
>> and the grading has not been processed so far for any student.
>>
>> So, we have the answers for randomized questions we currently do not
>> have. Is there any chance to create backwards a corresponding exam for
>> these questions with the CODE of the first exam which were meant for
>> the second exam? This would be okay in this case.
>>
>> My idea:
>> - Copy the course folder of the first exam
>> - Split the scanner results of the second exam by the two CODEs
>> - One Bubblesheet grading for the scanner results with the correct CODE
>> in the folder for the second exam
>> - Another Bubblesheet grading for the scanner results with the
>> "incorrect" CODE in the new folder
>>
>> I am not sure about the way how the CODEs and randomization are
>> created. Is it guaranteed that the copy of folder allows to use the
>> same CODE as used for the original folder? We need to make sure that
>> the randomization is fitting.
>>
>> Thanks for any help,
>> Stefan Bisitz
>> _______________________________________________
>> LON-CAPA-users mailing list
>> LON-CAPA-users at mail.lon-capa.org
>> http://mail.lon-capa.org/mailman/listinfo/lon-capa-users
>
> _______________________________________________
> LON-CAPA-users mailing list
> LON-CAPA-users at mail.lon-capa.org
> http://mail.lon-capa.org/mailman/listinfo/lon-capa-users
More information about the LON-CAPA-users
mailing list