Chapter 12 doubts

[C2H6O]

Hi, I have two questions.

1. There is line given on last page(72) which says that, grouping of sign test and lag-1 serial correlation test, test for the same thing.
Can anyone please explain it to me how?

2. In grouping of sign test, why do we only check for positive groups?? A more appropriate thing would be to first check which SD(positive or negative) has lower no. of groups and then test whether it is significantly small, as low no. of groups of any SD whether positive or negative would mean overgraduation.


Many thanks

 

[Tim.Sullivan]

Hi - I've just gone through ch12 myself. This is my view - please correct me if I'm wrong!

1. Both tests detect clumping of derivations of the same sign - I can't explain why in any more detail but I don't think we need to worry beyond that simple rule!

2. We're only interested in overgraduation - we don't usually use this test to detect undergraduation because we use the smoothness test for that. We are told if the rates are overgraduated, the deviations will not swap from positive to negative very often (and there will be fewer runs than expected). So we only need to be concerned to check that the number of groups of positive deviations are as many as we would expect.

Hope that helps

 

[C2H6O]

Hi, thanks for the reply.

For Question 1, my doubt still remains intact

As for question 2, I think you've misinterpreted the definition of overgraduation. Overgraduation doesn't mean only too many positive SDs. That's called overestimation. Too many negative SDs together would also mean overgraduation.

That's why I was wondering why do we only test for positive groups.
It's also given on page 47( last para) that, this test can lead to different conclusions depending on whether positive or negative groups were considered.

So my doubts still not cleared.

Anyone to help!