correspondence analysis

Lainaus Petra Wallem <pwallem at bio.puc.cl>:

Hi everybody, I am starting my friendship with R. Now I want to do a 
Correspondence Analysis using the vegan library or multiv library. Both give

me the same results which differ from the ones I get running other programs

like Statistica... I wounder if there is a problem with my matrix. I have 
cualitataive variable in the columns and sampling units in the rows... I 
import my matrix into R with matrix(scan....), which dose not accept strings,

so I have eliminated all the names of my rows and columns and just worked 
with the acctual values of my matrix, is that the problem I have?
Thanks a lot to everybody

There really is not enough information to help you. Further, I don't have
Statistica to see how it works. How were the results different? Scaling is not
quite standardized in CA, and it may be that the scores are linearly related
although they look different at the first glance. Further, some opt for using
square roots of eigenvalues. Checke these first. Actually, you should start with
checking your data. Do the dimensions match, or did you lose some rows or
columns in input? Do the numbers look OK? Please, be more detailed with your
problems.

From the description of your data matrix I would guess that you would rather

like to use correspondence analysis functions in MASS than those in vegan or
multiv. These all should give the same results, though, but they are geared to
different kind of problems.

Best wishes, jari oksanen