By the "null distribution" do you mean that the assignment of each
observation to a column is equal? If so, the function sample() might
serve your needs. For example:
rows <- 3
cols <- 4
rowtot <- 100
m <- matrix(NA, nrow=rows, ncol=cols)
for(i in seq(rows)) {
m[i, ] <- tabulate(sample(seq(cols), rowtot, replace=T))
}
m
[,1] [,2] [,3] [,4]
[1,] 27 24 25 24
[2,] 19 24 26 31
[3,] 26 26 31 17
Jean
Jim Silverton wrote on 08/29/2011 01:14:28 AM:>
> Hi all,
> I have a 3x4 contingency table with row totals all being 100. I want to
> generate 3 x 4 tables from the null distribution. Which R function can
do> this?
>
> --
> Thanks,
> Jim.
[[alternative HTML version deleted]]