Question about last weeks exam

I have a question about the graduation test question in Octobers Ct4 exam

It asked to do a normal goodness of fit test. However, it asked for 2 other statistical tests. Would you count the smoothness test as one such test?

The first one I did was the st dev test (as per normal). But from these results, it looked like the data was undergraduated.

Hence I did the smoothness (undergraduation) test as the other test?

Is this OK? What do you think?

 

I didn't do the smoothness test. Like you I didn't think this was a statistical test. But I kicked myself afterwards for not doing it. I suspect it would have failed the smooth test, because the grad rates were so close to the actual that it would have probably been an undergraduation.

It said statistical test didn't it? So I chose std dev and signs test. The graduation was ridiculously close to the data, spot on in many cases. For signs test the deviation was zero for many cases so don't know if that counts as a +ve or -ve or neither (think it says in notes somewhere but I can't remember).
And the chi-square was so small that it would not have failed the chi-square test as it was almost perfect fit.

I was disappointed in this question and ran out of time so didn't have time to think about it properly. That's one of my issues with the exams in general - it seems to be a desperate run against the clock to pick up as many marks as possible, another 5 minutes here and there to actually think more about the question would yield better answers.

 

General Opinion about Sep2009 CT4 Exam

Fellows what is your general opinion about CT4 exam of this session?

 

Hi,

I also did the signs test for this question and, like scarlets, wasn't too sure about what to do with the zero's. Time was an issue for me and so I just treated them as negatives and moved on...

I was wondering about that exposed to risk question where census data for the one year was missing (I think... don't know that I read it all that well). I decided to come back to this one at the end of the exam but when I did I didn't have enough time to answer it well. I started answering the question as though all data were present and stating what calculations would be made if this were so, and then went on to point out the fact that some amendment was required because data was missing and started stating asumptions... I never got to do much more than this. Does anyone know if this might earn you a mark or two??

 

Was I the only one who didn't have any zeros? You wern't comparing the ungraduated qx's with the number of deaths/exposed to risk are you? Then all you have is rounding errors. I thought the qx's in the paper were ungraduated, but maybe I miss-read the paper.

I thought you found the qx's from tables with the graduation formulia given (ie 10*qx I think)

I then multiplied these by exposed to risk and compared with number of deaths.

Am I way off the mark?

 

Yes, I agree with you, I didn't use the graduation formula but I realised that was what should have been done because those were the observed rates. Hope I get some marks for the method!