JDog, Thushan and pi, for what it's worth, I think the reason why losing marks on Maths Exam 2 is worse than losing marks on Maths Exam 1 is because Exam 2 tends to be harder, thus reducing the standard deviation (why exactly, I'm not sure), so each mark lost is a greater portion of the standard deviations lost.
If the two exams were of equal difficulty, we might see something like this:
Exam 1, Mean = 20, SD = 10
Exam 2, Mean = 40, SD = 20.
Now, each mark lost from Exam 1 is 0.1 of a SD, and each mark lost from Exam 2 is initially 0.05 of an SD. But Exam 2 is weighted double, so each mark lost turns out to be 0.1 of your final weighted score.
However, because Exam 2 is generally more difficult than Exam 1, we get something more like this:
Exam 1, Mean = 20, SD = 11
Exam 2, Mean = 40, SD = 18
So, each mark lost from Exam 1 is 1/11th of an SD, and each mark lost from Exam 2 is 1/18th of an SD, but is doubled to become 1/9th. Therefore, a person who gets 0/40 and 80/80 will do better than a person who gets 40/40 and 40/80. Note that in both these cases, the mean does not matter.
To give an actual example, I'll use my Methods scores from last year.
For Exam 1, I got 32/40, which was +0.95 SD's, multiplied by 0.22 (the weighting), to give 0.21
For Exam 2, I got 74/80, which was +1.78 SD's, multiplied by 0.44 to give 0.78. So, we'll say I have a final exam score of 0.99
Had I lost 6 more marks on Exam 1, and fullmarked Exam 2 (so still 14 marks lost in total), my final score would be around 0.22(0.42) + 0.44(2.10) = 1.017. Had I fullmarked Exam 1, and lost 8 more marks on Exam 2, my final score would be around 0.22(1.65) + 0.44(1.34) = 0.954.
Anyways, tl;dr, if you have to lose a certain number of marks, better to do it on an easier exam, where the standard deviation is higher (edited, I initially wrote lower), and each mark is worth 'less'.
Moderator action: removed real name, sorry for the inconvenience
Home
Help
Search
Login
