I have to do an analysis some stratified samples from the Survey of Income
and Program Participation. I am planning to use svyreg to estimate my
model. My question pertains to the specification of strata and PSU.
The documentation says, "In the public use files, analysts should look for
the following variable names for the variance stratum and variance unit
codes associated with each sample member: HHSC and HSTRAT in the core wave
files." HHSC is always 1 or 2, and HSTRAT is integer from 0 to 72. In
terms of Stata constructs, which is the psu and which is the strata?
I am pooling 12 samples covering two decades of time. During this time,
the survey design changed three times. My 1984 sample uses a design from
the 70s (seems to have 72 values for HSTRAT). Data from 1985-1993 use a
sample design from the 80s (seems to have 73 values for HSTRAT). data in
1996 uses a sample design from the 90s (seems to have 56 values for
HSTRAT). In pooling this data, do I need to create "new" PSU and strata
variables to account for the design changes? For example, I could add 100
to all strata in 1985-1993 and 1000 to strata in 1996. Should I create
this "interacted" PSU/strata by survey year instead (for example,
newstrata=strata+survey_year)?
Thanks in advance!
Regards,
--Alex Cavallo
Lexecon
(312) 322-0208 voice
(312) 322-0218 fax
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
