A 1-1 poly-t random variable generator with application to Monte Carlo integration

m-1 poly-t density functions have kernels which are ratios of a product of m multivariate-t density kernels to a single multivariate-t density kernel and arise as posterior densities for the coefficient of a single structural equation under a variety of assumptions. We propose a computer algorithm for drawing random samples from 1-1 poly-t distributions and discuss how to use these distributions for the evaluation of characteristics of higher order poly-t distributions. Illustrative examples demonstrate that this algorithm is fairly accurate at reasonable costs of computation.