Reputation: 3663
I was wondering if the cor.test function in R to compute Spearman's rho is tie-corrected or not. I couldn't find a straight answer in the official R documentation.
An example of tie corrected Spearman is found in:
Cureton, Edward E. The average spearman rank criterion correlation when ties are present. Psychometrika 23, no. 3 (September 1958): 271-272.
Thanks for any hint, Mulone
Upvotes: 3
Views: 5373
Reputation: 263362
The "official" documentation is the code itself. And looking there, one sees that there is provision for correction for ties through the use of pkendall()
.
stats:::cor.test.default
You will also get background information at this recent posting on SO regarding the Spearman-rho and three Kendall-tau's
Upvotes: 6