From: Bob Wheeler <bwheeler@echip.com>
Subject: Re: 4-Parameter distribution fitting
Date: Tue, 26 Jan 1999 21:35:46 -0500
Newsgroups: sci.stat.math,sci.stat.consult.,sci.math
To: arap@ms.com
Keywords: Fitting data to Johnson distributions

Johnson, N.L. (1949) Systems of frequency curves
generated by methods of translation. Biometrika.
36. 14-76. Most of this is also available in
Johnson and Kotz, Continuous univariate
distributions. Fortran software for fitting by
moments is available in statlib -- one of the
Applied statistics algorithms.

Moment fitting is basically unstable. Using
quantiles is better.  See
Wheeler, R.E. (1980) Quantile estimators of
Johnson curve parameters. Biometrika 67,725-8.

-- 
Bob Wheeler --- (Reply to: bwheeler@echip.com)
        ECHIP, Inc. 


Charles Pehlivanian wrote:
> 
> I am looking for references on 4-parameter distributions. The problem is
> to fit an empirical distribution to a particular instance of a
> distribution from a 4-parameter family, based on an estimate of the
> first four moments. I am aware of the so-called Johnson distributions,
> but do not have a solid reference.
> 
> Can anyone provide a refrerence for the Johnson distributions, or any
> general 4-parameter family that can be fitted to empirical data? Is
> there any general method along these lines? Any help would be
> appreciated.
> 
> Thank you,
> Ara Pehlivanian