On the sensitivity of sample L moments to sample size

Abstract

Parametric probability distributions can be fit to a dataset by equating sample L moments to those of the fitted distribution. This study examines the mean and mean squared departures of sample L moments of monthly precipitation data from large sample values as sample size increases. Mean departures decrease as the sample size increases with values near zero generally occurring with about 30 to 40 or more observations for the central tendency measure, about 40 to 50 or more for the dispersion measure, and about 60 to 70 for the skewness and kurtosis measures. -from Author