The Lognormal Distribution and the Translation Method: Description and Estimation Problems
André G. Laurent · Journal of the American Statistical Association · 1963
This paper studies the problem of variable scaling and recommends the use of “functional” characteristics as descriptive measures of central tendency and variability for a population (or probability distribution). It gives, in the case of a lognormal population, the minimum variance unbiased estimates of the median (geometric mean) and the logarithmic variance. Applications to time series are mentioned.