L-moments for data summary and distribution fitting

Speakers:  Nicholas J. Cox, Durham University, and Patrick Royston, Imperial College

The method of L-moments formalised by J.R.M. Hosking offers an elegant and useful means of summarising and comparing data sets. L-moments are more resistant to outliers, and for higher moments less biased, than classical estimators, yet are in many ways less ad hoc than several approaches based on order statistics. The method is also helpful for fitting various distributions to data, particularly when maximum likelihood is impractical. The method will be outlined with a variety of real examples, principally environmental and medical. The paper will introduce and explain Stata programs written by the authors.