Decomposition SE - IPUMS CPS

CPS
1

In Stata, I run svyset [iw=asecwt], sdrweight(repwtp1-repwtp160) vce(sdr) mse
oaxaca firm50_p $indepvars if sampMH==1, by(notcitizen) detail($vargroups2) svy categorical($categories).

However, I get the error message “svy sdr is not supported by suest.”

I tried svyset [iw=asecwt], sdrweight(repwtp1-repwtp160), and the decomposition runs, but I don’t get the correct SE.
I also tried, bootstrap, reps(1000): oaxaca firm50_p $indepvars if sampMH==1, by(notcitizen) detail($vargroups2) categorical($categories), but I didn’t get the correct mean differences or SE.

The main issue is obtaining the correct SE. Can you please point me in the right direction?

Thank you!

2

According to Stata’s documentation, the suest command works with results from the svy prefix command only for vce(linearized). Unfortunately I’m not aware of a solution to this incompatibility. Setting vce(sdr) is necessary to ensure that the standard errors estimated use the replicate weights. You may look for an alternative postestimation command to suest that can fit your needs and which is compatible with vce(sdr). I find the Stata forum can be helpful for these questions.

Please feel welcome to post again in the IPUMS user forum if you find a solution, since this might help other users.

1 Like