adjusting for sample design

[Pina Valle]

Hello.
I would like to analyze a subset of my data (teenagers). However, I am utilizing a command that does not accept the subpop option with svy. Therefore, I would like to adjust for weights, clustering, and strata and use the "if" option. I know how to adjust for weights and clustering without svyset, but is there any way to also adjust for strata? I also tried to use the "over()" option, but did not have any success. I am using Stata 15.
Thank you,
Pina

[Tiago Pereira]

Pina, hi. To provide you with guidance more easily, if possible, it is always a good strategy to describe the command you have been using and give us an example of dataset.

[Rich Goldstein]

I agree with Tiago Pereira - note that if you are using the "svy" prefix, subpop is an option to that prefix and thus is always available; we need to know exactly what you are telling Stata

[Pina Valle]

Thank you both for the reply. I am using “mvdcmp” -- this is a user-written program from ssc for a decomposition analysis.

When I run the following code:

mvdcmpDev groupvar: svylogit anyuse famst, subpop(age1519)

I get the following:

option subpop() not allowed
r(198);

[Rich Goldstein]

first, "mvdcmpDev" looks like a typo to me

second, -svylogit- was long ago replaced with -svy: logit-, and there the subpop option comes before the command; what version of Stata are you using?

I am not familiar with mvdcmp and do not know how 2 prefix commands should be used together; you might want to try tech support for this or the author of mvdcmp

[Pina Valle]

Thanks, Rich.

mvdcmpDev is correct -- it's an update to the original.

I am using Stata 15. I will check with the author since, to my knowledge, the updated command accepts -svylogit-

[Pina Valle]

Hello. I wanted to follow up on this post. I confirmed with the author that subpop cannot be used with this command. The "if" statement also cannot be used, though even if it could, documentation for the data I am using indicates it should not be used anyway.

Is there anything else I can do to run my analysis on a subpopulation without the subpop command?