Independence test for multidimensional data sets


I had a 6 dimensional (non parametric independent variables) data set, each case representing day of my study period. I performed clustering analysis and categorized the days into four groups. Now I am looking for a statistical test which can establish whether the 4 clusters are statistically independent or or not.

I can not do t-test or ANOVA as the 6 variables are non-parameteric.

I am not considering the The Kruskal-Wallis test as it can only handle single variable at a a time where as my data set has 6 variables.

Any suggestion in this regard?

Hey mzalikhan.

Have you considered running some kind of specialized regression and testing whether interaction terms are statistically significantly zero?

If you have a categorical data set you should be able to do this in a major package.