Dear Statalisters,

I have Stata 15 and have recently started using LCA. I have been analysing a set of 6 binary variables and am using the following code.


Code:
gsem (bmi disease sex age_cat height_cat weight_cat  <- _cons), logit lclass(C 3)
However when I get to testing the GOF of the model, I only get 0.000. From the examples I have read, the p-value that they get for the likelihhod ratio means that they fail to reject the null hypothesis that their model fits as well as a saturated model but they still go ahead with the analysis. I have attached a screenshot of the example.



Code:
estat lcgof

----------------------------------------------------------------------------
Fit statistic        |      Value   Description
---------------------+------------------------------------------------------
Likelihood ratio     |
       chi2_ms(1002) |   2621.749   model vs. saturated
            p > chi2 |      0.000
---------------------+------------------------------------------------------
Information criteria |
                 AIC | 55028.330   Akaike's information criterion
                 BIC | 55188.835   Bayesian information criterion
----------------------------------------------------------------------------
I wanted to ask if in choosing the best model, we should also take into account the p-value or just stick to the BIC.

Apologies for such a basic question.

Thank you.