Best correlation approach for likert scale data

noetsi

Fortran must die
#1
I have a set of four point likert scale measures (highly satisfied, somewhat satisfied, not satisfied, highly dissatisfied). I have to run correlations between these. Last time I looked at polychoric correlations and spearman rho, I was wonder which of these is better for this type of data or if there is another preferred alternative?


Another concern, there are a lot more satisfied than dissatisfied people, I have never read that this is an issue with correlations, but I figured I ask.
 
Last edited:

noetsi

Fortran must die
#2
I get a strange error when I separate the analysis by area (this does not show up when I run all the data together). It says

The number of groups for variable Q1 is less than 2, the affected polychoric correlations are set to missing.

Note sure what that means. Strangely it will generate Spearman's rho with this problem.