Correlate nominal/metric data with ordinal data?

I am currently finishing my bachelors thesis and have a couple of questions regarding correlation in spss. I have a questionnaire that generates metric data, nominal data and ordinal data. I used Spearman to correlate all the ordinal datasets and I am quite comfortable that this is fine. Now comes the hard part: How do I correlate the metric data with my ordinal data? Metric data consists of age and kilomenters driven per year. Ordinal data consists of "agree", "agree partly", "neither agree nor disagree", "disagree partly", "disagree". Is it possible to turn my metric data into ordinal data with 3 values (e.g. "young", "middle aged", "old" and then use spearmen to correlate it to the other ordinal data with spearman?

Additionally I would like to ask how I can correlate the ordinal data with the nominal data (being "yes" and "no")? Unfortunately I don't have any idea on how to do that.

• – ttnphns Jul 14 '18 at 18:40

Spearman can be used for the correlation of metric data and ordinal data. The function will convert the metric data to ranks. You don't want to break your metric data into three ordinal categories as you suggest.

For an ordinal and a nominal variable, you can use Freeman's theta or epsilon-squared. Search this site for "Freeman's theta" for additional discussion.

In the case of having only two levels in nominal variable, there are other options. You could use Spearman correlation for this too.† Or use Cliff's delta. In the case of two levels in the nominal variable Cliff's delta will equal Freeman's theta.

† You would want to correlate Y and Group, not Group1 and Group2.