Hi all,

My query is regarding fixed-effects in stratified survey data.

I am working with cross-section, firm-level data that specifies both probability weights and the strata variable. So I begin with:
-svyset id [pweight=wt], strata(strata),

where the .dta file indicates that
gen strata = group(country location industry)

Now I was reading a paper that uses the same dataset for a straightforward probit regression, but the analysis also includes country-specific effects as well as industry-specific effects. My question is: Wouldn't setting "-svyset id [pweight=wt], strata(strata)" where "gen strata = group (country location industry)" take care of these country-specific or industry-specific effects? Or do these effects do need to be added separately even after the survey design has been set to include strata that was formed on the basis of these characteristics only?

Or is it possible are the authors only specifying -
probit y x1 x2 i.country i.industry, [pweight=wt] and are ignoring the strata? Wouldn't that be an incorrect approach?

I wrote to the authors regarding this but they say that they only use the "standard probit model from Stata 14.", not sure if this helps me.

Any advice, please? Thanks!