I would much appreciate your help with the following problem:
I am not able to produce predictive margins for gsem with random effects. One example is that I want to show how the changing of proportion of foreign population influences probability of voting for populists.
Dataset:
Dependent variable has 3 categories - (1) voting mainstream parties, (2) populists, (3) non-voters.
Due to fractions of missing data I am using multiple imputation by chained equations.
full model:
Code:
mi estimate, dots cmdok: gsem (i.volba <- i.man migranti_2011_bezS migrantiS_squared vek_rek vek_squared_rescale i.religiozita i.bydliste ses i.vzdelani_3k egalitarstvi lidr i.amm egal euroskepticismus_rek apatie cynismus spk fear M1[okres_cd]@1 i.amm#M2[okres_cd]@1, mlogit) [pweight=vaha], technique (nr bhhh dfp bfgs) difficult
Code:
mimrgns, at(migrantiS_squared=(0(1)45)) predict(outcome(2.volba)) cmdmargins vsquish marginsplot, noci scheme(sj) name(mimrgnsplot1) mimrgns, at(migrantiS_squared=(0(1)45)) predict(outcome(2.volba) mu fixedonly) cmdmargins vsquish marginsplot, noci scheme(sj) name(mimrgnsplot1)
only one of mu or fixedonly is allowed
Can you advice how to modify my command? Thank you in advance.
0 Response to Computing predictive margins (mimrgns) for gsem with random effects
Post a Comment