[R] fisher.test - FEXACT error 7

[Bernardo Rangel Tura]

On Fri, 2009-03-20 at 18:29 +0000, Helena Mouriño wrote:
> Dear all,
> 
> Im having an awkward problem in R.  When I write the command
> Fisher.test(school.data,workspace=2e+07), where school.data is the matrix
> corresponding to the data set,
> I get the error  message:

> FEXACT error 7.
> LDSTP is too small for this problem.
> Try increasing the size of the workspace.
> 
> Increasing the workspace: 
> Fisher.test(school.data,workspace=1e+10),
> I get a different message, but it still doesnt work:
> 
> NAs in foreign function call (arg 10)
> In addition: Warning message:
> In fisher.test(dados, workspace = 1e+10, alternative = "two.sided") :
>   NAs introduced by coercion

Hi Helena,

In this case you can try 3 solutions:

1- chisq.test(school.data), but pay attention if expected value of any
cell is < 5 

2- Fisher.test(school.data,workspace=2e+07,hybrid=TRUE) from Help

For larger than  2 by 2 tables and 'hybrid = TRUE', asymptotic
     chi-squared probabilities are only used if the 'Cochran
     conditions' are satisfied, that is if no cell has count zero, and
     more than 80% of the cells have counts at least 5.

3- Use "large tables" approach from Sir David Cox:

Law, G. R. and Cox, D. R. and Machonochie, N. E. S. and Simpson, J. and
Roman, E. and Carpenter, L. M. (2001) Large tables. Biostatistics
2(2):pp. 163-171.


-- 
Bernardo Rangel Tura, M.D,MPH,Ph.D
National Institute of Cardiology
Brazil