Help System (web edition)

T2TEST <data1>,<data2>,<output_line> 
compares mean vectors of two multivariate samples given as Survo
data <data1> and <data2>.
By default T2TEST performs Hotelling's T2 test thus assuming
multivariate normality and population covariance matrices to be equal.
It also continues by making a randomization test by using the T2 test
statistics without multivariate normality assumption.
The number of replicates in the simulation process is given by
a SIMUMAX specification. Default is SIMUMAX=10000.

When a specification METHOD=2 is given, T2TEST performs Yao's T2 test
assuming multivariate normality but allowing unequal covariance
matrices (generalized Behrens-Fisher problem).
See e.g. "Methods of Multivariate Statistics" by M.S.Srivastava (2002).
Also in this case the process is continued by a randomization test
using the T2 statistics without assumption of multivariate normality.
   M = Comparing multivariate samples