Two-step estimation of multivariate polychoric correlation

Abstract
In this paper, a two-step procedure for estimating multivariate poiychoric currelacions is investigated. Both the generalized least squares and maximum likelihood methods are considered. Asymptotic properties of the two-step generalized least squares estimator are established, and its relations with the two-step maximum likelihood estimator are studied. Iterative procedures are developed to produce the estimates and the standard error estimates. An example is provided to illustrate the theory.