topic Re: P value in Discussions

P value

kowa — Mon, 02 Sep 2024 09:39:54 GMT

Bonjour, je suis noovice sur JMP et j'aimerais comprendre Pourquoi JMP ne donne pas de p-value pour certaines distributions comme une loi normale d'ordre 2 ou 3?

Merci

Re: P value

MRB3855 — Mon, 02 Sep 2024 10:21:06 GMT

Hi @kowa : Welcome to the community! I'm not sure what you are asking. Do you mean to ask why there is no Goodness-of-fit Test for the "Normal 2 Mixture Distribution" or "Normal 3 Mixture Distribution" in the Distribution platform?

Re: P value

kowa — Mon, 02 Sep 2024 15:10:55 GMT

Hi @MRB3855 ,

yes, that's my question.

Indeed when I have data adjustments that give me a lognormal distribution, I can have the P-value for Goodness-of-fit but jmp does not propose this if I have a normal distribution of order 2.

Re: P value

MRB3855 — Mon, 02 Sep 2024 16:13:56 GMT

Hi @kowa : JMP has no GOF test for these mixture models. And there are good reasons for this. See here:

https://community.jmp.com/t5/Discussions/Goodness-of-Fit/td-p/229235

For a more complex explanation/discussion/alternative see here:

https://www.sciencedirect.com/science/article/am/pii/S0167947318301142

And here:

https://www.martinmodrak.cz/2018/05/14/identifying-non-identifiability/#:~:text=What%20is%20%E2%80%9Cnon%2Didentifiability%E2%80%9D,probability%20distribution%20of%20observed%20data.