hi all
I wish to estimate the regional difference in the effect of transfers on government popularity in a country with basically only 2 regions (Belgium)
if I add to my regression specification not only legislature (= time) fixed effects but also legislature fixed effects interacted with a region fixed effect, my coefficients of interest change considerably
why would this be?
does this suggest there are time invariant but region and legislature specific factors associated with both transfers and government popularity?
so would this be my preferred specification?
however I have only 215 observations (= 43 subregions * 6 legislatures)
ould adding 6 more variables (due to the addition of a legislature fixed effects interacted with a region fixed effect) be more than I can afford?
cheers, Geert
I wish to estimate the regional difference in the effect of transfers on government popularity in a country with basically only 2 regions (Belgium)
if I add to my regression specification not only legislature (= time) fixed effects but also legislature fixed effects interacted with a region fixed effect, my coefficients of interest change considerably
why would this be?
does this suggest there are time invariant but region and legislature specific factors associated with both transfers and government popularity?
so would this be my preferred specification?
however I have only 215 observations (= 43 subregions * 6 legislatures)
ould adding 6 more variables (due to the addition of a legislature fixed effects interacted with a region fixed effect) be more than I can afford?
cheers, Geert
