On Fri, 11 Feb 2005 07:22:13 -0600, Laurel Copeland
<[email protected]> wrote:
> According to the SIPP Manual
> http://www.census.gov/apsd/techdoc/sipp/sipp96l.pdf page 8-1:
>
> "The 1996 Panel of the SIPP sample is located in 322 Primary Sampling Units
> (PSUs), each consisting of a county or a group of contiguous counties.
> Within these PSUs, living quarters
> (LQs) were systematically selected from lists of addresses...."
>
> As far as I can tell, there is no SIPP variable for the PSU. The PSU code
> is scrambled inside the ssuid variable (the household ID number) but I do
> not think there is any way to tell which ssuids came from the same PSU.
>
> However, from reading the Stata 8 manual at U [30] p. 346-347 I wonder if I
> should use the command:
>
> svyset wpfinwgt, strata(gvarstr) psu(ssuid) fpc(epppnum)
No, this one is wrong on both psu and fpc variables.
Indeed, it looks like the Census people are fascinated with the
generalized variance function approach to estimation rather than
anything else (Statistics Canada are fascinated with the bootstrap,
for instance :)), and hiding the PSUs for confidentiality reasons. You
would have to use their Table 4 for your computations... and I don't
know what to do with more complex stuff like regressions. You indeed
would need to find somebody at the federal office to explain the
intricacies of the standard error computation.
Check also that fpc, the finite population correction, is used when
you are sampling more than 10% of the population, so that there is a
difference between the with replacement (easier, more conservative)
and without replacement (more difficult, and at any rate Stata seems
to track only the top stage of selection) formulae. It is interpreted
as the population size, and Stata will be puzzled to find it varying.
--
Stas Kolenikov
http://stas.kolenikov.name
*
* 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/
