How to report Correspondence Analysis

Adri

New Member
#1
Hi.
I'm busy with a scientific research honours thesis, and for my results the statistician at my university conducted a correspondence analysis with a scatterplot, and then also a table. The problem is, I have no idea how to report this in my write-up. I have looked on the internet again, and know I'm supposed to report the chi-value. Problem is, I can't find it in my stats. What I have is the Dimensions (Dimension 1 on x-axis, Dimension 2 on y-axis), inertia % and Eigenvalue.
The problem is, I have no idea how to write these results up in my report..
Statistica was used to do the stats.
Thanks in advance for any help!
 

gianmarco

TS Contributor
#2
Hi!
I am more than happy to help you on CA since this exploratory technique is not so widespread among this forum's user :)
I do not want to complicate things too much, but what should be reported along with CA scatterplots depends on the way(s) in which the analysis has been conceived by the statistician you were referring to.

I believe that at least you should report:
1) the original contingency table
2) the row and column categories' contribution to the definition of the dimensions keept for the analysis
3) the correlation of row and column categories with the selected dimensions

It would be too long to illustrate the rationale of (2) and (3), since it would mean to delve into the core of CA underlaying logic.
So, I would prefer to point out to you an article of mine (see this link) where an example can be found.
For the CA in general, I have just put on-line a website on CA (at this link).

Please, feel free to write back for any other question.

Hope this helps for the moment,
Regards
Gm
 

Adri

New Member
#3
Thanks so much! That example really helped a lot already! I will contact you, should any further questions arise. You're a bit of a life-saver.