On 5 Mar 2003 17:05:04 -0800, [EMAIL PROTECTED] (dennis roberts) wrote:

> here is a scenario that you might want to explore with some of your 
> students ...
> 
> say that you have given what we might consider a typical 75 item 
> multiple-choice test ... exam is given from 4PM to 5PM ...
> 
> what you are interested in is the relationship between the time they take 
> to finish (ie, turn it in) the test and their scores on the test
[ ... ]

I accidentally took part in something *half-way*  like this.
In an undergraduate course in sociology, circa 1967, 
the professor commented that he had an observation to
pass on, concerning the multiple-choice test he was 
returning;  "we might want to think about it," he said.
 - To this day, in long retrospect, I don't remember enough
details to be able to figure out whether the guy was teaching
us a lesson, cleverly; or had a great sense of humor, or just what.

Unbeknownst to us, we had taken two versions of the test,
with the a-to-d  answers arranged in different orders.  
He had created version A (let's call it) by randomizing the
responses in his head, while inventing the test.  He 
subsequently created version B  by re-arranging the order
of the version A.    [ Yes, if he were a psychologist, he 
should have known this was a biasing procedure.  Humans
do not do a good job in creating random order, especially
on that first task. ]

While taking the exam, we had taken the time we wanted
within the hour, and we were allowed to submit the
test and leave the classroom whenever we finished.
His casual observation, on returning the exams, was 
that the first 7 completed had been version A, and the 
last 10,  all version B.   - and that worried him that 
(essentially) the tests might not be strictly parallel.  
Therefore, he was afraid to compare the resulting grades; 
so he hadn't.  



Doesn't any well-designed multiple choice test lend itself
to having a 'speed'  component that will correlate with
better 'intelligence', better preparation, better outcome?
You might get the opposite result if you put in a just a
couple of items where only the best students will take 
a lot of time.  Why would that be?  Obscure calculations?
I wonder, could that still be a well-designed test? - Yes,
I think so, if the questions are arranged with increasing
difficulty.



-- 
Rich Ulrich, [EMAIL PROTECTED]
http://www.pitt.edu/~wpilib/index.html
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to