Dear Statalisters
Thank you very much for your attention to my post.
I employed a discrete choice experiment and am currently analyzing the data using the mixlogit command in Stata. My hypotheses state that the respective effects of IVs on DV (i.e., choice) vary by participants' disability status (i.e., d_by_round==0 means people without disabilities; d_by_round==1 means people with disabilities). Therefore, I run exactly the same mixlogit models twice, just each time with different subsamples as below.
[Stata commands that I use]
global randvars "permanent parttime zerohour home hybrid flexible"
mixlogit selectedchoice att5_wage if d_by_round==0, rand($randvars) group(gid) id(pid) nrep(500)
mixlogit selectedchoice att5_wage if d_by_round==1, rand($randvars) group(gid) id(pid) nrep(500)
To compare the coefficients from models 1 and 2, I ran the "estimates store A" command after the first mixlogit command and the "estimates store B" command after the second mixlogit command. Up until this part, Stata did not give me any error message.
However, as I typed the "suest A B" command, Stata says,
"unable to generate scores for model A
suest requires that predict allow the score option
r(322);"
Can anyone kindly let me know how I can compare the coefficients of the two mixlogit models which use different subsamples in Stata?
Since I have many IVs with which I expect to have disparate effects on DV, I could not specify interaction terms, and, hence, comparing the respective mean coefficient values was pretty much the only way that I could test my hypotheses.
Your help would be deeply appreciated.
Very respectfully,
Thank you very much for your attention to my post.
I employed a discrete choice experiment and am currently analyzing the data using the mixlogit command in Stata. My hypotheses state that the respective effects of IVs on DV (i.e., choice) vary by participants' disability status (i.e., d_by_round==0 means people without disabilities; d_by_round==1 means people with disabilities). Therefore, I run exactly the same mixlogit models twice, just each time with different subsamples as below.
[Stata commands that I use]
global randvars "permanent parttime zerohour home hybrid flexible"
mixlogit selectedchoice att5_wage if d_by_round==0, rand($randvars) group(gid) id(pid) nrep(500)
mixlogit selectedchoice att5_wage if d_by_round==1, rand($randvars) group(gid) id(pid) nrep(500)
To compare the coefficients from models 1 and 2, I ran the "estimates store A" command after the first mixlogit command and the "estimates store B" command after the second mixlogit command. Up until this part, Stata did not give me any error message.
However, as I typed the "suest A B" command, Stata says,
"unable to generate scores for model A
suest requires that predict allow the score option
r(322);"
Can anyone kindly let me know how I can compare the coefficients of the two mixlogit models which use different subsamples in Stata?
Since I have many IVs with which I expect to have disparate effects on DV, I could not specify interaction terms, and, hence, comparing the respective mean coefficient values was pretty much the only way that I could test my hypotheses.
Your help would be deeply appreciated.
Very respectfully,
Comment