Studies in Support of the Application of Statistical Theory to Design and Evaluation of Operational Tests. Annex A. A Methodology for Determining the Power of Manova when the Observations are Serially Correlated.
Abstract
This research addresses two related problems of multivariate statistical analysis. First, the effects of a multivariate time series on the MANOVA power function are investigated through the use of an experimental design. Second, a generalized procedure is developed for incorporating the multivariate time series into the MANOVA power function so that the effectiveness of ANOVA and MANOVA models in evaluating command and control systems, on the basis of powers of the tests, may be made. In order to make the analysis possible, a procedure to determine the power of MANOVA test was required. The MANOVA power function is not known in a closed or usable form; consequently, a Monte Carlo procedure was used to determine the power of the MANOVA test. The maximum likelihood form of the MANOVA test statistic was utilized due to its ease of computation.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1975
- Accession Number
- ADA052074
Entities
People
- Norviel Robert Eyrich
Organizations
- Georgia Tech