Kendall's Tau or Spearman's Rho?

Question

I am interested in the effect of an ordinal variable (likert type) that had 4 answer options on participants preferences of six statements (as participants ranked these, I assume this too would be ordinal data).

In the case of my data, would kendall's tau or spearmans rho be more appropriate. I know Kendall's tau can be used with tied ranks, but I am unsure as to what that means in relation to my data. I know there is a similar thread, but it doesn't have any data examples, so it is hard for me to compare it to my own.

Any advice would be much appreciated

It is interesting that the question's title is almost exactly the same as the earlier duplicate. — Michael R. Chernick, Apr 05 '17 at 10:09

score 1 · Accepted Answer · answered Apr 05 '17 at 10:40

1

I would suggest that Spearman's test is a safe option and the way to go in your case.

Both test are non-parametric so no assumption about the underlying distribution of your data. However, I believe that in general Kendall’s Tau-b is more appropriate for square tables which is not your case.

answered Apr 05 '17 at 10:40

Vasilis Vasileiou

1,158
1
9
16

One of the Kendall's $\tau$s may handle ties better and be easier to interpret for this case. – Frank Harrell Apr 05 '17 at 11:16

Kendall's Tau or Spearman's Rho?

1 Answers1