[R] Cochran-Armitage

[mailbox@fontaine.net]

Hi,
 
I have to calculate a Cochran-Armitage test on data. I don't know much 
on this test and then, I have few questions on the independence_test 
function. I have read the help and search the archives but this doesn't 
help me...
 
- They are two ways: either with a formula (independence_test(tumor ~ 
dose, data = lungtumor, teststat = "quad")) or with a table 
(independence_test(table(lungtumor$dose, lungtumor$tumor), teststat = 
"quad")).
Why this difference ? The two ways do not give the same results... Why 
?
 
- In the teststat option, what is the difference between "quad", "max" 
and "scalar" ? What is used generaly ?
 
- Why choosing approximate or asymptotic distribution if an exact test 
can be calculate ?
 
For a classical Cochran Armitage test (nothing special was specified 
so I guess I have to do the most classical Cochran Armitage test), 
which options should I use ?
 
Thanks for your help.
 
Delphine Fontaine