At 01:02 PM 2/10/2003, jim clark wrote:
I think my point was ... though, probably it could have been stated more clearly ... that IF the adjustment of cut scores was rationalized because of measurement error ONLY ... then, there is just as much in that argument to say move the cut UPward (which we dare not do) ... as downward (which we do fairly often) ...I'm not sure I fully appreciate the distinction between "measurement error" and "not perfect" or "tolerance."
Measurement error would imply that what we see as an observed score ... could be in error either way and, cut scores by that justification could be about as likely moved UP as well as down.
My own opinion is that usually, when we move cut scores DOWN ... to allow more to have the NEXT higher grade ... we are mainly doing that to compensate for the fact that we don't, as instructors, have any really good justification for the grade we give ... literally done according to some preset scale ... and, to get us out of this potential (end of semester) bind we tend (if we do it at all) adjust the cut scores so that students are not as able to argue that our original scale was bad ... since we can point to the adjusted scale AND the students and say ... see, I am taking that into consideration AND IN your favor.
We know that if we adjusted the scale UPward ... and now cut out some students from getting a B- ... according to the up front scale ... and now get C+ ... we are going to be in for a real donnybrook of a fight ... and maybe even a legal challenge too.
Thus, it is not so much giving the benefit of the doubt TO the student BEcuase of measurement error, it's that we want to make it LESS likely that we will be challenged on the grade that we give.
That makes our life easier ... and generally makes the student happier too.
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
