Estimation of Parameters of a Multivariate Normal Population from Truncated and Censored Samples

Abstract

SUMMARY This paper is concerned with the problem of estimation of means and variances of a k-variate normal distribution from a sample doubly truncated or censored on s-variates (s ≤ k), when variates are uncorrelated. Maximum likelihood equations are derived which, with the help of normal probability tables, can readily be solved by iteration. The information matrix is given from which the asymptotic variances and covariances of the estimates can be determined. to illustrate the practical application of these results a numerical example is added.