No recommendations for reading, but a few thoughts....
Course/exam difficulty is utterly confounded with (concerted) student
investment of effort, so if you use a scaling system, it shouldn't
depend only on the marks.
Think about the scale on which the ultimate mark are to be reported -
typically criterion-referenced marks ranges, e.g. with 40/50/60/70
borderlines. Not an interval scale, or a ratio scale? What
transformations would be legitimate?
... and if you know a way to set a statistics exam so that a bare first
class performance yields only 1.75 times the marks of a bare third
performance, you may not need to scale your marks (and please let us all
know how to do it!)
Peter
|