[Lon-capa-users] Grading

Mark Lucas lon-capa-users@mail.lon-capa.org
Mon, 11 Feb 2002 16:12:15 -0500 (EST)


	We just went through a little round of handgrading. As much as I
have been trying to run interference, the other instructor decided to try
to do some of it by herself. Fortunately, the hardware is still intact.

	Most of this has probably already been mentioned, but their may be
added insights here. One caveat is that we are using the stable release as
of about middle January, I think.

	We realize there are more options with this system - in fact, we
ended up re-opening the assignment for the two students and logging in as
the students to enter the answers. This was still possible since I
assigned local passwords and they can't change them online yet 8)

	It would be really nice for the instructor to be able to view a
completely active page or sequence as that student would see it, down to
the ability to enter answers and all. I could imagine that there is a
pandora's box opened there, but I haven't thought about it that carefully

	I think the key here is that information is spread around a lot of
different places - you have to look at problems as individual entities,
rather than a page; you have to evaluate answers under submissions, then
go to grades under grading.

	We need a way to view a sequence for a particular student with
full handgrading capabilities incorporated. In addition, the answer needs
to have tolerance information available side-by-side with the answer in
order for the instructor to evaluate the correctness of the answer.

	The current submissions information for a single problem does not
include any tolerance information.

	In examining the submissions output, the instructor was getting
flustered by the part numbers. Because we use the include files from CAPA,
our first part ID is typically 14, second 16, .... Part of this is
training. She was totally unaware what the parts meant. Part of this was
real - these numbers are meaningless to a typical person who has not been
staring at the source code of the problem. On basically has to count on
the sequence rather than the names. I'm not sure if there's a good scheme
for working this - especially with the flexibility created by the system.

	It's also becoming clear to me that userids aren't much better
than student ID numbers. We either need a convenient name lookup system
for userids, or more options to choose a student by name directly.

	Any thoughts on the above?


Mark Lucas					email: lucasm@ohiou.edu
252D Clippinger Lab  				phone: (740)597-2984
Department of Physics and Astronomy             fax:   (740)593-0433
Ohio University
Athens, OH 45701