Allen--
I don't think the HH/person weights you refer to are a problem, in
that the HH weights are (I think) the person weights of the HH head,
or some random adult in the HH. Since HH characteristics that you
might tabulate at the HH level (i.e. each HH is represented as one
record when you compute a mean) are attached to person records (in the
standard one-to-many mapping) when you merge based on HH serial
number, using the -merge- command, the person weight is all you need
for the "combined" analysis.
There are cases where you have multiple distinct weights, e.g. if you
have sample weights (inverse of the probability of selection for
sample members) and precision weights (inverse of the standard error
on the estimate of some number for that sample member), in which case
I believe researchers usually multiply weights together. That would
not be appropriate for the HH level/person level analysis you are
talking about, I think.
--Austin
ps. data referred to is freely avail at http://www.ipums.org
On 8/2/06, Allen Glicksman <[email protected]> wrote:
I was wondering if there is any way to combine different types of
weights in the same analysis. I have already contacted Stata and read
the manuals, and am assuming the answer is no, but just in case I
thought I would ask the list.
I have a couple of data sets that include individual and household
variables. For example, in IPUMS (the individual level Census sample)
there are individual level variables (country of birth) and household
variables (linguistic isolation). There are separate weights for
individual level variables and household level variables as well. If
I wanted to use both variables in a regression analysis for example,
is there a way of using the weights for each variable?
Thanks,
Allen Glicksman
*
* 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/
