Title: Multivariate mean testing under block exchangeable covariance structure
Authors: Ivan Zezula - P.J. Safarik University (Slovakia) [presenting]
Daniel Klein - P.J. Safarik University (Slovakia)
Anuradha Roy - The University of Texas at San Antonio (United States)
Abstract: A multivariate linear normal model with block exchangeable covariance structure is considered. Such a structure can be naturally implied by the design of the experiment. Its main advantage is strongly reduced number of the second order parameters, which allows mean testing even with a small sample size. We will compare performance of several test statistics used in this setting. Methods will be demonstrated on real dataset.