I have two ordinal variables which are set on different scales. The variables were extracted from a questionnaire. The first variable ranges from -12 to +12 with a step of 1 width, ( -12, -11,...-2,-1 , 0, 1, 2,..., +11, +12) and the second one ranges only from -2 to +2 with the same step width(-2,-1,0,+1,+2). I would like to investigate whether they correlate and also measure the strength of a potential association between them. How do I do this? Do I have to standardise the variables in order to compare them to each other? Thank you in advance, Konstantinos

  • 2
    $\begingroup$ Paying attention to how wide is the step width exposes that one actually thinks his scales are interval, not ordinal. Ordinal scales do not have a defined step width, albeit they do differ in respect to the number of levels (=categories, =points): some are more grained and some are less. $\endgroup$ – ttnphns Jan 4 '13 at 9:50
  • $\begingroup$ hi, thank you for your comment. Still, how would these variables be compared to each other? $\endgroup$ – Konstantinos Jan 4 '13 at 10:14

Polychoric correlations were developed for precisely this problem. There are implementations in R.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.