1
$\begingroup$

I have read up Somers' D and everywhere it states that the statistic is a measure of association between ordinal variables. Can someone explain to me why I cannot use this metric for assessing the relationship between an ordinal or binary dependent variable and a continuous independent variable? Assuming I have a finite sample, it should always be mathematically possible to assign ranks to a continuous variable and calculate Somers' D. In other words, is Somers' D still applicable, if my dependent variable continuous?

$\endgroup$
1
  • 2
    $\begingroup$ continuous variables are ordinal $\endgroup$
    – JTH
    Jun 27, 2021 at 18:35

1 Answer 1

1
$\begingroup$

It is perfectly fine to use Somers' D if one or even both variables are continuous.

The only requirement is that values can be compared within variable.

Depending on which of the two variables is continuous, D will correspond to Kendall's tau.

$\endgroup$

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.