I often use the reduced chi squared as a quick goodness of fit test when fitting histograms. Is there an analogous method that works more generally for interpreting the absolute value of the best fit likelihood when data are not normal? I'm thinking of binomial and poisson with noise here but the more general the better. I guess what I might be fishing for is an answer to this question.
$\begingroup$
$\endgroup$
Add a comment
|