|
|
Comparing models with PP checking? |
|
Message/Author |
|
|
Dear professors, I am running CFA:s using Bayesian two-level models, and I wonder if there is any way to compare model fit between these models? Because these are two-level models, the only model fit information is posterior predictive checking, and I wonder if this can be used somehow to compare models? Of course, if one model is clearly off (significant p-value) and another is not, I guess that's information supporting the latter model, but can we say for instance that a model with a p-value of .20 has better fit than a model with a p-value of .10? Best, Fredrik Falkenström |
|
|
No direct way to do this. It would have to be done for each MCMC draw it seems. |
|
Back to top |
|
|