Testing of hypothesis of a structured mean vector for multi-level multivariate data with structured correlations on repeated measurements.

ANURADHA ROY; RICARDO LEIVA

Salt Lake City, Utah, USA

Congreso; Joint Statistical Meeting,(JSM 2007); 2007

In this article we study testing of hypotheses of structured mean vectors for three- level multivariate data with Kronecker product struturered covariance matrices for multiple q-variate over u-sites and over p-time/spatial points under the assumption of multivariate normality. We provide the maximum likelihood estimates of the unknown population parameters for both the autoregressive of order one and compound simmetry correlation structures of the repeated measurements over time. We also provide the computation algorithms to calculate the test statistics. The tests are implemented with four real data sets.