An analysis of a multivariate two-way model with interaction and no replication
In this paper, we consider the problem of testing the hypothesis of no interaction in the multivariate two-way classification model with one observation per cell. If the error covariance matrix is unknown, we suggest pseudo-maximum likelihood estimation (PMLE) of the parameters and give the invariant test for the hypothesis of no interaction. Further, we develop also some tests for the case of an unknown diagonal positive definite error covariance matrix.
Key wordsAnalysis of variance interaction pseudo-maximum likelihood estimate maximal invariant invariant test
Unable to display preview. Download preview PDF.
- Guo Dawei. Tests for the Hypothesis of a Class of No Interaction in the Multivariate Two-way Classification Model.Acta Scientiarum Naturalium Universitatis Jilinesis, 1983, 3: 15–27.Google Scholar
- Krishnaiah, P.R. Computations of Some Multivariate Distributions. In Handbook of Statistics 1, North-Holland, Amsterdam, 1980.Google Scholar
- Rao, C.R. Linear Statistical Inference and its Applications. 2nd Ed., John Wiley & Sons, 1973.Google Scholar
- Krishnaiah, P.R. On Generalized Multivariate Gamma Type Distribution and Their Applications in Reliability. Proc. Conf. Theory and Appl. of Reliability with Emphasis on Bayesian and Nonparametric Methods (I.N. Shimi and C.P. Tsokos, eds.), Academic Press, New York, 1977.Google Scholar