Last 3 Days
Latent Variable Mixture Modeling
posted on Sunday, January 01, 2017 - 9:44 pm
I¡¯m trying to do an analysis of bayesian using the mixture IRT 2PL model.
My questions are:
1. Regarding how to input ¡®prior¡¯:
1-1 I found a way to input prior for ¡®loading¡¯ and ¡®threshold¡¯, but I¡¯m wondering, how I can get the prior for ¡®discrimination¡¯ and ¡®difficulty¡¯? Can you let me know?
1-2 Given that the inverse Gamma (-1,0) for ¡®f(latent variables)¡¯is prior, which is default, how can I change this into a ¡®normal distribution¡¯?
1-3 With the given value, 'prior', for 'c(latent class)', how can I change this into ¡°c~bernuolli (¥ğ1,¥ğ2), (¥ğ1,¥ğ2)~dirichlet(0.3, 0.7)¡±?
2. How can I get the DIC or posterior mean of the deviance?
Unfortunately, my English is not so great, so I'm hoping that any answer you can provide for my questions below use only syntax, if possible, please.
Thank you in advance!
Bengt O. Muthen
posted on Monday, January 02, 2017 - 2:30 pm
1-1 Because discrimination and difficulty are not the parameters estimated, you cannot give priors for them.
1-2 Gamma is used for variances. I don't think a normal prior is suitable for a variance.
1-3 I can't understand your notation here.
2. If DIC is not printed it means that it has not yet been developed for this model.
Back to top
Add Your Message Here
This is a private posting area. Only registered users and moderators may post messages here.
Enable HTML code in message
Automatically activate URLs in message