Re: SPSS - two non-dichotomous categorical variables

Posted by Art Kendall on Sep 23, 2014; 5:16pm
URL: http://spssx-discussion.165.s1.nabble.com/SPSS-two-non-dichotomous-categorical-variables-tp5727364p5727367.html

If you are still in the design stage of you study I strongly suggest that you use more levels in your Ordinal/interval variables.  Since the underlying construct is continuous this would be a better operationalization of the construct.

If you are already have the data, start by doing a scatterplot of your variables.

If your levels are ordered your variables are at least ordinal.

ask for all statistics on CROSSTABS.  The output groups the coefficients by what they are useful for.

http://pic.dhe.ibm.com/infocenter/spssstat/v22r0m0/index.jsp?topic=%2Fcom.ibm.spss.statistics.help%2Fspss%2Fbase%2Fidh_xtab_statistics.htm

describes the coefficients.  

You would be interested in predicting your DV from your IV.

Are your results very different from assuming that the levels are interval, i.e., that the the intervals between the values are not very discrepant?  If you have CATREG available in your installation of SPSS you can actually compare the fits.