SVYSET for pretest and posttest survey

[Sonam Rinchen]

Dear Sir/Madam,

I am writing to seek your guidance on an issue I've encountered regarding SVYSET and weighting.

I have successfully merged the pretest and posttest datasets, where the pretest dataset has 13,336 samples and the posttest dataset has 2000 samples. Each dataset comes with its respective weight variable - "weight" for the pretest and "weight_r2" for the posttest.

To conduct survey-weighted analyses, I have set up the SVYSET as follows:

1. For pretest: svyset unique_draw [pw=weight], strata(stratum44)
2. For posttest: svyset unique_draw [pw=weight_r2], strata(stratum44)

However, when performing t-tests or regressions using SVYSET on variables like "hhmem" (pretest household size) and "b1" (posttest household size), I have observed that both variables seem to adopt the SVYSET of the posttest, rather than adhering to their respective SVYSET specifications.

I would greatly appreciate your insights on how to address this issue. Specifically, I am seeking guidance on how to ensure that each variable follows its designated SVYSET specifications, given the unique weighting for each dataset.

Your expertise in this matter would be invaluable, and I am grateful for any assistance you can provide. Thank you very much for your time and consideration.


Thanking you