Estimation for Lognormal Data with Multiplicative Errors

Abstract

Actual data C’ are noisy versions of the true process C. Pollution data is often assumed lognormally distributed with multiplicative errors. Given this, the use of appropriate stochastic models to describe C and C’ is discussed, together with procedures for making statistical estimates about the parameters of C. The usual normal procedures produce biased results. The one way balanced ANOVA design is analyzed and applied to two data sets.