0

I had the same question as in this post: Factor Analysis with factomineR

It partially answered my question, but I also wonder how do we interpret such results.

I know that when we have variables with more than two levels each, we can use the graph to tell if the variables are over- or under-represented. However, in this case since we cannot have the graph.

I want to know the relation between the variable RJT_MEM_STG (2 levels) and the OAL_SRC_IND (6 levels).

This is my code

RJT<-c(361,2719,0,12016,11,23734)
NON_RJT<-c(0,1125,1,867,1,5161)
mat<-cbind(RJT,NON_RJT)
rownames(mat)<-c("B","E","F","J","M","T")
mat<-as.matrix(mat)
res_ca<-CA(mat, graph=F)
res_ca$row$coord
res_ca$row$contrib
res_ca$col$coord
res_ca$col$contrib

I get the following output:

Output for the rows

Output for the columns

I know that we should interpret only the varaibles which have the bigger contributions. So, I focus on the variable J and 1.

Should we do the rest of the analysis with the coordinates ?

Can someone help please ?

KoenV
  • 4,113
  • 2
  • 23
  • 38
alixletouze
  • 11
  • 2
  • There is no point of doing a CA on a 4*2 table. CA are made to reduce the dimensionality of **large** contingency table. If your variables have so few values, you better just interpret the contingency table directly, using chisquare or fisher test if needed. – scoa May 24 '17 at 07:53
  • Okay, I was not sure if the Chi-square test was as interesting as the CA. But in fact, I will do a ChiSquare test ! Thank you for your answer! – alixletouze May 24 '17 at 11:59

0 Answers0