st: Discrete time duration model with heterogeneity: xtcloglog and pgmhaz
I am running a discrete time duration model including unobserved heterogeneity.
I tried to run it with xtcloglog (assuming Normal distributed errors) and with
pgmhaz (Gamma heterogeneity).
The thing that worries me is that I am getting very different results,
especially for some selected covariates. Particularly, some variables that have
significant coefficient with xtcloglog are not significant with pgmhaz.
I wonder how I could choose the right model for my data.