1
$\begingroup$

I have a couple of variables with different scales - I don't want to use contingency tables so Is there a way to calculate all the correlations with different methods and then compare them?

$\endgroup$
2
  • 1
    $\begingroup$ Are the variables quantitative, qualitative? $\endgroup$
    – Drew75
    Feb 7, 2014 at 18:08
  • $\begingroup$ both - unfortunately $\endgroup$
    – Lara
    Feb 7, 2014 at 18:41

1 Answer 1

1
$\begingroup$

You'll have to use different methods to compare correlations. Unfortunately for the cases with qualitative variables, you'll need contingency tables.

  1. Quantitative-quantitative: Easy! Spearmean, pearson, and kendall correlation (scales don't matter).

  2. Quantitative-qualitative: Less easy. You can use a t-test to check the equality of means of the quantitative variable over the modes of the qualitative variable. Or analysis of variance. These are contingency-tables.

  3. Qualitative-qualitative: Less easy. This is a Chi-2 test. Period.

You won't be able to escape contingency tables.

$\endgroup$
1
  • $\begingroup$ @Lara Welcome to Cross Validated :) If the above answer helped, please accept so others know you got the information yuou needed. $\endgroup$
    – user31668
    Feb 8, 2014 at 1:23

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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