hi everybody,

Its been a years way back since I used the correspondence analysis (CA) as a statistical tool. I need advice as to how could I run the analysis using SPSS but the data has three columns - one for the brand, one for attribute and the other for the weights (percentages). I understand that in SPSS there is only row and column to specify and that the CA is only for nominal data? How would I run my data so I can get as to what attribute a particular brand is related? And how would I arrive at the plot of these variables?

Please advice.

Melisa

Hi Melissa,

what you need to do here is multiple correpondence analysis. The CA you find in soos under analyze-data reduction-correpondence analysis is simple CA. If you choose optimal scaling in stead of CA, you can choose all variables are nominal and then you can run MCA.
I know this is the way it works for spss version 14, I also know that this does not work with version 12..so it does depend on which version of spss you have if you can run the MCA.
I hope this info is of some use for you.

hi everybody,

Its been a years way back since I used the correspondence analysis (CA) as a statistical tool. I need advice as to how could I run the analysis using SPSS but the data has three columns - one for the brand, one for attribute and the other for the weights (percentages). I understand that in SPSS there is only row and column to specify and that the CA is only for nominal data? How would I run my data so I can get as to what attribute a particular brand is related? And how would I arrive at the plot of these variables?

Please advice.

Melisa

Very interesting. I would to know if one has to dichotomize (0,1) nominal variables before running MCA in SPSS 14. Thank you.


Christian

Dear colleagues, :wave:
I've never used correspondence analysis but now I have to use it in my work. Please, suggest me some links to articles or books in Internet. Or if you have any materials about it in elecronic version and can share it with me, send me them on my mail kleinLSD@*****.com. :confused: It's very urgent.
Thanks in advance,
Miroslava

dear colleagues
you can see this links for understanding correspondence analyses:
http://www2.chass.ncsu.edu/garson/pa765/correspond_spss.htm
i think is easy to understand