A few comments on the subject of grade distributions:
1. I first encountered bimodal distributions about 15 years ago (and
I've been seeing them ever since). I commented about such a distribution
in exam scores for engineering physics to a chemistry faculty member at
that university (in Wisconsin), and he said he had been seeing such
distributions for quite some time. He attributed it to the fact that
a generation earlier, the people who were in the bottom part of the
distribution probably would not have been admitted to college.
2. 'Motivation' is something that is very important in understanding
grades. For about 4 years, I've been giving a simple math pretest to
the students in my intro physics classes. The test is only about
12 or 13 questions long and covers arithmetic, simple algebra, geometry,
and has one trig problem. It is given to the students without any
advance warning, so it gives information about what they _really_ know
how to do. I've been looking for any correlation between the pretes
score and the final grade in the class. There is no strong correlation
unless two groups of students are excluded from the analysis:
1. Students who score high on the pretest and do poorly
in the physics course
2. Students who score low on the pretest and do well
in the physics course
To me, these obviously represent 'motivational' effects. (By the way,
group 1 is much larger than group 2...) When these groups are excluded,
there is a small positive correlation between pretest score and final physics
grade.
Steve Luzader
Frostburg State University
sluzader@fre.fsu.umd.edu