[R] Anderson-Darling gof test (was "no subject")

[Rolf Turner]

On 03/10/14 03:54, eliza botto wrote:
> Dear UseRs,
> I obtained following results from Anderson-Darling Goodness of fit test.
>> dput(EB)
> structure(c(2.911, 0.9329, 0.818, 1.539, 0.604, 0.5142, 0.4344, 0.801, 0.963, 0.9925, 0.933, 0.956, 0.883, 0.572), .Dim = c(7L, 2L), .Dimnames = list(c("EXP", "GUM", "GENLOG", "GENPARETO", "GEV", "LN", "PAR3"), c("A2", "P")))
> My question is how to interpret these results? More precisely, which distribution fitted well on my data?


(1) Please use an informative subject line.

(2) Your posted data are somewhat opaque; one might guess that "A2" 
holds the values of the test statistic and "P" the corresponding 
p-values.  However I cannot reproduce your "P" column by applying the 
pAD() function.

(3) I don't really know anything about the Anderson-Darling test, but 
the function ad.test() requires that parameters of the distribution be 
supplied.  What is the impact of these parameters' being estimated, as 
they must surely be?

(4) This is really a statistics question rather than an R question, and 
so is not appropriate for this list.  Moreover it would appear that you 
are well out of your statistical depth.  You should seek local 
statistical consultation rather than trying to scramble your way through 
topics of which you have no proper understanding by peppering the r-help 
list with ill-posed questions.

cheers,

Rolf Turner


-- 
Rolf Turner
Technical Editor ANZJS