MULTIVARIATE PAIRED COMPARISONS: SOME LARGE-SAMPLE RESULTS ON ESTIMATION AND TESTS OF EQUALITY OF PREFERENCE

The paper is a second one on multivariate paired comparisons. The first paper contained the methodology and this paper contains some large-sample results on estimation and test of preference equality. The limiting normal distribution is derived for the joint distribution of estimators for attribute preferences and correlations; the dispersion matrix derived indicates desired asymptotic variances and covariances. The likelihood ratio test of equality of preference parameters has the noncentral chi-square limiting distribution and its noncentrality parameter is given under local alternatives. Asymptotic efficiencies of the method of paired comparisons relative to a multiple multivariate sign test are computed. Report on The Development of Statistical Methods for Quality Control and Surveillance Testing. See also FSU Statistics Rept. M161, AD-685 698.