In an ideal world, the marking schemes used in laboratories would produce a range of marks, which when combined with some students not completing all exercises, would lead to a range of overall marks with an acceptable mean average and standard deviation. Current wisdom suggests a good mean average is 65%. In practice, most course unit managers construct marking schemes for each exercise which award almost full marks to all students who complete the work! This leads to unacceptably high averages for laboratory work, which distort the overall examination results presented in the examiners' meetings. ARCADE counters this `evil' with a lesser evil: the scaling factor. As the marks come in, ARCADE dynamically computes a scaling factor for each course unit so that the mean average (before late mark truncation) lies between 60% and 65%. (Exceptionally, some course units have a different range, or are unscaled.) Usually this means marks are scaled down to obtain a 65% average, but they can also be scaled up to obtain an average of 60% - a useful safeguard if marking schemes turn out to be too strict, or a course unit suffers unexpected severe problems.
Scaling factors are always linear, and are always a whole percentage.
The scaling factor for each course unit only gets fixed once all expected work is completed and marked - typically shortly after the laboratories are finished. This means that a student cannot reliably predict his or her marks prior to this, although the scaling factor in use near the end of the semester is unlikely to be very different from the final one. A positive side effect of this is that students cannot stop working the moment they know they have passed.
The downside is that some students complain that they are unable to obtain 100% in their laboratories, even though they hand in ``perfect'' work. Phrases like ``get a life'' should be sensitively and sympathetically passed to such people, plus the important view that nobody, not even them, is perfect!