Fisher exact test with zeros

[ekbrown77]

I have a 5 X 6 contingency table, with zeros in 14 of the 30 cells:

Given that the expected values of most of the cells is smaller than 5, I try to run a Fisher exact text (instead of a chi-square test), but get the following error:

How should I go about determining statistical significance with regards to independence of these two categorical variables?

[Stefan Th. Gries]

Works here:

fisher.test(qwe <- matrix(c(0,1,1,0,4,0,1,0,0,5,0,57,28,1,50,0,3,0,0,2,0,1,0,0,0,1,14,4,0,5), ncol=6))

Maybe the problem is that your input is a data frame?

Cheers,
STG
--
Stefan Th. Gries
----------------------------------
Univ. of California, Santa Barbara
http://tinyurl.com/stgries
----------------------------------

[ekbrown77]

Well, that's embarrassing. Yeah, that was the problem. Thanks.

I just discovered data.matrix(), which converts from type data.frame to matrix.