Quant help please

[simski81]

OK, so mild panic has set in and i dont think I'll get to look at quant again now.

Could someone please set out the following:

1. when to use a t-test for mean of differences

2. when to use a f-test for difference of mean

3. is this the only thing we use the f-test for?

4. Chi-x is only used for testing whether the variance of two populations are the same.

If someone can answer the above I will love you forever and would bear your children if i could.

 

[bchadwick]

Use a t-test for mean of differences if your sample N<20 or so. Otherwise, the z-score is a sufficient approximation.

I don't think you use an F-test for differences of means (though maybe I'm not considering some situation it was used). The F-test is usually used to test whether a regression has any validity, asking "what's the probability that all coefficients are = 0." There are other ways of using the F test and I don't remember them all, but don't think they are tested at L1.

Chi^2 is used for lots of things - mostly about deciding if the difference between a set of outcomes and a set of expected outcomes are statistically different from each other or not. At Level 1, the main use is testing whether variances are the same or not.


Hope that helps... although there are a few people who would object if you bore any of my children. Thanks for the offer, though. ;-)



Edited 1 time(s). Last edit at Monday, December 1, 2008 at 06:26PM by bchadwick.

 

[jlive1975]

I wouldn't spend too much time on this topic. At the high end, I'd guess there would be 2 questions on it tops. You're better off trying to understand a FSA topic or Ethics/GIPS in more detail at this point.

Check out my blog

http://myCFAadventure.blogspot.com