# Nonparametric ICC

Anyone can suggest me a nonparametric version of the intraclass correlation coefficient (ICC)? My model have the following details: two-way random effects model, consistency, average ratings as unit. I am using irr R-cran package.

thank you

-
The ICC is not, in itself, a parametric statistic. You can conceive of the ICC regardless of the distribution of the data. What exactly do you mean by "nonparametric ICC"? –  Macro May 25 '12 at 14:51
I have seen that in some studies the ratings are firstly examined to check a possible violations from normality, and then ICC is computed. Is it mandatory? What if the ratings are far from normality? –  netrunner May 25 '12 at 15:17