Interval Estimation for the Coefficient of Variation for Lognormal Population with Added Zero Values

摘要

This paper considers the coefficient of variation for Iognormal population with added zero values. Specifically,we are interested in interval estimation. We propose the exact approach and the generalized pivotal approach. The simulation results indicate that when proportion of zero observations are large the generalized pivotal approach has better than the exact approach. The coverage probabilities of the generalized pivotal approach are generally close to the nominal level and the average lengths are smaller than that of the exact approach.