Ali Abdi wrote:
Dear allstat members
I have two random vectors, and I want to test whether these two vectors
are independent or not (each vector contains non-independent but
identically distributed random variables).
I'm familiar with the standard tests for that purpose, as described in
standard multivariate analysis texts like Anderson or Muirhead. However, I
cannot use them because of the assumption of normality. My two random
vectors are highly non-normal. So basically I'm looking for a
non-parametric test.
I also appreciate if somebody introduce me a software which can
non-parametrically test the independence of two non-normal random vectors.
Thanks in advance,
Ali Abdi
Ali,
I think you can use the Mann-Whitney test and the (U) and the
Kolmogorov-Smirnov Z.
Regards,
Glauber Fonseca
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|