Hi all,
I am quite new to Stata and I am trying to run a Bernoulli-logit regression with gsem builder. I have a latent independent variable and a dichotomous dependent variable. The independent variables are personal values construct, while the independent is the support for populist parties. So far, I got good results for three out of four of the latent construct. With the last one, I obtain unexpectedly high coefficients. This is the command I am using:
gsem (open -> ipcrtiv, ) (open -> impdiff, ) (open -> ipgdtim, ) (open -> impfree, ) (open -> ipadvnt, ) (open -> impfun, ) (open -> populistvotebase, family(bernoulli) link(logit)), latent(open ) nocapslatent

And these are the results I obtained:


Array
All the other led to results like the following (the results are coherent with the distribution of the independent variable, going from 1-6):

Array

Does anyone have suggestions on why I obtain such different results for the "open" construct?
I already check for the construct to be stable enough (I previously ran a CFA with the latent constructs and they all show good fit).

Thank you in advance