Exam Standards

There was an article in the latest issue of 'The Actuary' on one of the questions in the CT exam on Statistical Models.
It showed the question set and the model answer required by the exam board and showed where the author of the article disagreed strongly with the ambiguity in the question from a Mathematical point of view. It mentioned 'Independant and Indentically distributed' twice in the question. The first time it meant Pair-wise, the second time it didn't.

Coming to the question directly from the notes and with no mathematical back-ground I can see how people were able to make the correct assumption as to what the Examiners had in mind. But having studied Mathematics as a degree the words have a more reigorous meaning to me (and the author of the article). Like him I would have got the question wrong simply because I would have assumed that the question was *not* ambiguous, and therefore its proper mathematical interpretation is the correct one.

It worried me that I could potentially be thwarted in an exam by having *more* rigorous understanding of Mathematics.
Is anything being done to prevent this situation occuring in the future, or are we supposed to treat every Mathematical statement in the exams in a 'looser' sense?

Thank you,
leafy

 

Following on from what I posted before, I came across this in ActEd's own notes that I think sums up what I was talking about perfectly.

If you have it to hand: Page 22 of Chapter 14, Analysis of Variance, for subject CT3, Probability and Mathematical Statistics.

I quote:

So what this is effectively saying is that the natural, and logical approach for any mathematician attempting this particular question is to use a one-tail test. However, for some reason that is not explained, in the exam we would be required to use a two-tailed test. So unless you specifically remember this case having a more advanced knowledge of mathematics actually counts against you!
I don't understand why the examiners are allowed to make up their own laws of mathematics.

If anyone else comes across situations like this could they please post them here so that no-one gets caught out by a situation such as this!

leafy

 

No, I agree with you. (I'm a mathematician)

The test H0 was: mu2 = mu3.

Since the means are arranged in ascending order the alternative, and thus the H1, should be: mu2 < mu3.
This is because it is clear to see that mu2 <= mu3 in all situations.

No test is "wrong", but clearly allowing for a situation that we know to be impossible is foolish and redundant.
A one-sided test is the most appropriate to use for the exact reason that we have one-sided tests for situations like these.

Why must I forget everything that I know about mathematics to be able to correctly answer an Actuarial maths question?

leafy

 

Could you PM me a contact e-mail or similar?

 

I sent an email at the beginning of the week and am awaiting the reply.
I will post any reply I get here (as long as I get the permission of the person who replies to me) so that everyone can benefit.

leafy

 

I think that covered most things actually...although the one/two-sided issue ball is apparently in ActEd's court...

leafy

 

Not entirely related, and I'm not sure if this is a Trevor Watkins thing or a Jane Curtis (as chair of it) thing, but the minutes of the last Student Consultative Committee are a lot duller than previous ones. In particular they don't list all the comments that came in, so you can't see any more if you were alone in commenting on something. I guess it also means that it's harder for people to follow up on comments that were made to see if anything happened...