Page 1 of 1

Correspondence Analysis

Posted: Mon May 01, 2023 9:44 am
by Beth
Hi,

I'm trying to run a correspondence analysis on Jamovi. I followed an example (https://sebastien-le.github.io/medasite/CA.html) I found online, and even though my data seems appropriate for the analysis, I get an error message "non convenient data".
I added the nominal variable as the row labels and the scale variables as active columns.
I also tried changing the data setup but I still got the same error. Can someone tell me what I'm doing wrong, please?

Re: Correspondence Analysis

Posted: Tue May 02, 2023 10:41 am
by seol
Hi Beth

Have you tried 'snowCluster' module for correspondence analysis?

Best

Seol

Re: Correspondence Analysis

Posted: Wed May 03, 2023 8:28 pm
by Beth
seol wrote: Tue May 02, 2023 10:41 am Hi Beth

Have you tried 'snowCluster' module for correspondence analysis?

Best

Seol
Hi Seol,

I installed and tried it after your suggestion. I'm very new to data analysis so excuse me if I'm missing something very obvious but I'm a bit confused about this. I'm getting "the following variables are not quantitative" warning for the nominal variable section. I have a summary table data as seen on the screenshot. I did the analysis on SPSS, but I guess the data setup must be different in Jamovi? I'd appreciate it a lot if you could explain a bit.

Thanks for your help,
Beth

Re: Correspondence Analysis

Posted: Wed May 03, 2023 9:54 pm
by seol
Hi

I think you need at least 3 nominal variables to run. you can try it using example data within snowClustermodule; Open>Data Library> Multivariate analysis.

Best Regards

Seol

Re: Correspondence Analysis

Posted: Thu May 04, 2023 10:21 am
by Beth
seol wrote: Wed May 03, 2023 9:54 pm Hi

I think you need at least 3 nominal variables to run. you can try it using example data within snowClustermodule; Open>Data Library> Multivariate analysis.

Best Regards

Seol
Hi Seol,

Thanks for your suggestion, I checked the example and it turns out the frequencies need to be nominal variables instead of scale. Once I converted them the analysis worked.

I still couldn't figure out why the MEDA module correspondence analysis doesn't work because my data looks exactly like the example but I get the "non-convenient" warning. I did the analysis on snowcluster but I'd still appreciate it if anyone can explain why this is the case.

Best,
Beth

Re: Correspondence Analysis

Posted: Fri May 05, 2023 7:15 am
by sle
Hi Beth,
you have to put at least 2 active columns.
If you could send me the data I can check, but it should work fine.