Skip to content
Prev 14527 / 20628 Next

mixed mutlinomial regression for count data with, overdisperion & zero-inflation

On 18/05/2016 08:26, St?phanie P?riquet wrote:
Stephanie,

Make a histogram of the 1000 values of the percentages of zeros....and 
present the 69.3% as a big blue/red dot. If the dot for your observed 
data is in the tails you have a problem.

I don't see the point of a test in your case. Such a simulation is close 
to bootstrapping...so I guess you can come up with a test somehow. If 
you do this type of analysis in a Bayesian framework it is often (and 
confusingly) called a Bayesian p-value (counting how often the simulated 
value is larger than your observed one).

I would just go for the histogram...seems you are lucky.

Alain