Apologies for long description. Quick search did not reveal the answer, hopefully will get directed into the right direction.
I am trying to develop questions that take mean ± error.
First, the mean is checked. Second the error value is verified. Third step is to verify that number of significant figures is one in the error or two if first significant figure is one. In the last step the number of decimal places is matched between the error and the mean. This part is working more or less.
Implementing penalties proven to be trickier.
It looks like currently STACK questions subtract the penalty let say 10% for each try. If a student performs badly, scores 25% on the first attempt and does not improve within 2 tries the score becomes 5%.
Is it possible to change penalty behaviour that 10% penalty is applied to the final score via the following expression: maxmark*(100%-10%)^Ntries ? Motivation behind is not to heavily penalise lower scoring students, while keeping roughly the same penalty to top marks.
With regards, Viktor