[R] two sample chi-squared test

[David Winsemius]

On Jan 29, 2010, at 11:11 AM, eric lee wrote:

> Hello,
>
> Can you tell me what R function to use to do a two-sample chi-squared
> test?  I want to see if two distributions are significantly different
> from each other, and I don't specify the theoretical distribution of
> either.  For example, I have the following fake count data:
>
> x <- sample(1:10,50,replace=TRUE)
> y <- sample(1:10,100,replace=TRUE)
>
> I saw chisq.test in the stats package, but that looks like a
> one-sample test.

Nope. You do need to supply it with the right organization of data,  
though. In your case that would be a 2 x 10 or a 10 x 2 matrix (so  
(2-1)x(10-1)=9 d.f)

?table
?matrix


> I'm running version 2.7.2 on windows xp.

Probably has very little bearing on this problem, but that is  
considered to be badly outdated in 2010.

> Thanks.
--

David Winsemius, MD
Heritage Laboratories
West Hartford, CT