Hello everyone
I am running a multinomial logit model using gsem
I tried finding marginal effect of x1 using the command:
margins, dydx(x1) predict (outcome(2.y) fixed)
where x1 is the independent variable of interest and y1 is categorical variable with 3 categories.
In the command above i wished to find marginal effect of x1, when x1 has some effect on second category of y , with first category as the base
the command seemed to have worked fine.
but could someone clarify if it is correct ?
because the sign of marginal effect is coming opposite to the sign of the coefficient of x1 on 2.y, with first category as the base
i would be really grateful if someone could reply in this regard
thanks in advance