by "gunnar" <gunnar@[EMAIL PROTECTED]
>
Feb 22, 2006 at 03:43 PM
I'm with Herman on this. When I started setting up freely available
on-line quizzes which students could take anywhere and anytime, someone
pointed out that it would be easy to set up a computer which would weed
through them and find all the correct answers. The trick in that case
was to have lots of these questions include random numbers. Of course
simply having a large enough database of questions would also solve the
problem since it is eventually more tedious to look up the right
question rather than work out the solution.
My typical exams are always (pretty much) the same from year to year.
I also tell this to my students and I show them all the earlier exams.
The trick is that each question is phrased so that it requires
understanding of the problem and all I do is to change the wording in
minor ways, switch around items and change numbers. As a result I
always get grades spanning the entire scale even though the exam is
always pretty much the same. For example I never state that the student
is supposed to use a certain t-test. Rather, I give data and ask
whether e.g. a statement is justified. Next year I will change the
statement or the data or... In class, when prepping for the test I
show the students how I can generate these questions...
I do rather enjoy those sessions when I show the students how one can
sensibly test their knowledge:-)
I've been able to do this in college/grad school mathematics,
statistics and fishery science which I teach - I don't see why it
shouldn't be possible in most subjects to focus on questions which
require understanding - and to always come up with (slightly) new ones.
Student who try to learn the answers to last year's exam will fail.
Those who understand last year's exam will pass.
G