I am trying to fit an mlogit model as shown below. Using the margins post command, I am also trying to estimate margins of responses for specified values of covariates. For some reasons, I would like the margins estimates to be zero for some of the levels of my outcome variable. I achieved this by defining constrains, however the margins command still returns non-zero margins estimates for the levels for which I set the coeff to be zero. I think this is happening because the constant for those levels for which the constraints are set to be zero are not zero. Thanks for you help on this:
*set coeff to be zero)
constraint 1 [14]
constraint 2 [15]
constraint 3 [25]
constraint 4 [41]
constraint 5 [51]
constraint 6 [52]
mlogit obesity_t c.age_100##c.age_100 covid##i.race i.raeduc_r , baseoutcome(11) constraints (1 2 3 4 5 6)
margins raeduc_r#race, at (age_100=(0 (1) 25)) atmeans force nose saving(tran_point_F_, replace )
constraint 1 [14]
constraint 2 [15]
constraint 3 [25]
constraint 4 [41]
constraint 5 [51]
constraint 6 [52]
mlogit obesity_t c.age_100##c.age_100 covid##i.race i.raeduc_r , baseoutcome(11) constraints (1 2 3 4 5 6)
margins raeduc_r#race, at (age_100=(0 (1) 25)) atmeans force nose saving(tran_point_F_, replace )
