I have some survey results which I have compiled into a crosstabulation (9x4)using SPSS.

I was intending to use chi squared analysis to determine if the results are significant, but in my stats notes (completed this unit some time ago....) I have that in order for chi squared to be appropriate the expected values should not be "too small". Many of the entries in my table are zero.

Can I use chi squared or should I use some other analysis method?

Thanks for your help

Regards

Michael