On Wed, 1 Mar 2000, Ajit K Jena wrote:
I am facing a peculiar problem and hope someone out there can comment on it. In goodness-of-fit tests for evaluation of distributions, there are three well-known methods: 1. Chi-square 2. Anderson-Darling 3. Kolmogorov-Sminrov
I would suggest you find the likelihood or some modified version of it (AIC). If the model describes data well, the likelihood is big. Of course you don't get a p-value, which in my book is a good thing. (Read the manifesto: AWF Edwards (1972/1992), Likelihood, Johns Hopkins U Press)
I certainly would agree with that. That is essentially what all my libraries do. For example, gnlr and gnlr3 in library gnlm when used with a null model (no covariates) will allow the comparison of about 25 different distributions. You might also like to take a look at Chapter 4 of my Introductory Statistics (OUP, 1995) which evaluates goodness of fit of a variety of distributions when the data are in grouped frequency form, and the accompanying R function, fit.dist in library gnlm (and for more advance likelihood inference, my Parametric Statistical Inference, OUP, 1996). Jim
Bill
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._