calculation of RRR from regression coefficient after latent class analysis using multinomial family

[Mili Dutta]

I was performing latent class analysis using multinomial family. But it produced regression coefficients but I want Relative Risk Ratio (RRR) as I am using multinomial. Please tell me how to produce RRR after performing LCA in STATA.

the syntax I have used is

gsem ( aa bb cc dd ee ff gg hh ii jj kk <- _cons) (A <- i.place_residence i.religion i.age_group i.ageatmarriage ) [iweight = wgt], family(multinomial) link(logit) lclass(A 4)

[Roman Mostazir]

All you need is to exponentiate the coefficients. Type the following code after executing your model:

estat eform

[Weiwen Ng]

I was performing latent class analysis using multinomial family. But it produced regression coefficients but I want Relative Risk Ratio (RRR) as I am using multinomial. Please tell me how to produce RRR after performing LCA in STATA.

the syntax I have used is

gsem ( aa bb cc dd ee ff gg hh ii jj kk <- _cons) (A <- i.place_residence i.religion i.age_group i.ageatmarriage ) [iweight = wgt], family(multinomial) link(logit) lclass(A 4)

Mili,

One clarification. I think most people would call this a latent class regression. You are saying that your variables aa-kk vary over 4 latent classes, and separately, fit a multinomial regression model for the latent class as the dependent variable and the other variables as covariates.

You did not need to specify -family(multinomial)- for Stata to recognize that A is multinomial. By doing so, I believe you are telling Stata to treat aa-kk as categorical random variables. Make sure this is what you want to do.

[Mili Dutta]

Thank you so much all for reply. My issue has been resolved.