Abstract
Modified maximum likelihood estimators of the parameters of a bivariate normal distribution are developed when the smallest and largest observations on one of the two components are censored. These estimators are shown to be unbiased and efficient for large n. For testing the null hypothesis statistic is defined.