I have a question related to Mark Fackrell's (see below).
I have been using the Kolmorgorov-Smirnov test to assess
the goodness-of-fit of parameters which were estimated
from the data using maximum likelihood.
I am aware that this constitutes an intrinsic hypothesis,
and have used the critical values published by L. H.
Miller (J. Am. Stat. Assoc. 51,111-121,1956).
This table seems to reject values which, by visual
inspection, appear to be a very good fit to the data.
How can I obtain a null distribution which is appropriate
to my problem?
Many thanks
Oliver Pybus
[My probability model represents waiting times in a
non-homogenous death process].
>I was wondering if anyone could help me out.
>
>I have fitted a distribution to some continuous empirical data.
>
>Could you please tell me some of the ways that I can assess/test the
>goodness of fit of the distribution to the data?
>
>Thanks
>
>Mark Fackrell
-------------------------------
[log in to unmask]
-------------------------------
University of Oxford
Department of Zoology
South Parks Road
(01865) 271272
-------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|