Chosing between different Models.

So I have a data set, and my job currently is to divine which model
describes the data the 'best'. I know, I know, tough call. This is
exponential decay data, given a particular distribution of states. Now
several distributions fit the data with almost equivalent chi-squares
and number of parameters. Is there any way of straightening out which
one is the most likely?

Note, I have many many measurements (~500) with varying environments -
and so the the distributions are effected as well, so if that will help
my search any. However, I would love to go back and correlate these
changes to something useful.

.

Relevant Pages

  • Re: Calculation of FWHM
    ... The distributions of RGB luminances are rarely Gaussian ... data set of dervatives, Laplacians, wavelet detail coefficient, etc., then ...
    (sci.image.processing)
  • Re: Chosing between different Models.
    ... I know, I know, tough call. ... exponential decay data, given a particular distribution of states. ... several distributions fit the data with almost equivalent chi-squares ... and so the the distributions are effected as well, ...
    (sci.stat.edu)