Dear all,
I would like to compare marginal effects from two panel multinomial logit models (xtmlogit) for two different subsamples (Macro==0 and Macro==1). A simplified versions of my models are as follows:
More precisely, I am using a panel multinomial logit model (xtmlogit), regressing a categorical migration variable (0=decrease, 1=no effect, 2= increase) on a factor variable weather (0= precipitation, 1=temperature, 2=drought, 3=disaster), factor variable development (0=low, 1=medium, 2=high) and a binary variable women. I study two different subsamples: Macro==1 and Macro==0.
Is there a way to a) compare the coefficients of marginal effects of al variables between the two models, and b) compare whether the two models systematically differ with one metric?
Thank you so much.
Best,
Barbora
I would like to compare marginal effects from two panel multinomial logit models (xtmlogit) for two different subsamples (Macro==0 and Macro==1). A simplified versions of my models are as follows:
Code:
xtmlogit Migration i.weather i.development i.women if Macro==1, vce(cluster ID)
eststo multinomial_macro
foreach o in 1 2 3 {
quietly margins, dydx(*) predict(outcome(`o')) post
eststo multinomial_macro_`o', title(Outcome `o')
estimates restore multinomial_macro
}
Code:
xtmlogit Migration i.weather i.development i.women if Macro==0, vce(cluster ID)
eststo multinomial_nonmacro
foreach o in 1 2 3 {
quietly margins, dydx(*) predict(outcome(`o')) post
eststo multinomial_nonmacro_`o', title(Outcome `o')
estimates restore multinomial_nonmacro
}
Is there a way to a) compare the coefficients of marginal effects of al variables between the two models, and b) compare whether the two models systematically differ with one metric?
Thank you so much.
Best,
Barbora
