[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: clustered data and two weights
I am using data from several European countries (ESS2) and want to use
robust standard errors AND the two weights that comes with the dataset.
One weight takes account of differences in sample design between
countries (individuals or households for example) and the other takes
account of the fact that differences in sample sizes do not correspond
to differences in population sizes between countries (sorry if I'm not
using the right technical language here...). To use both weighs, I have
multiplied them to get one "combined" weight, as you can only use one
weigh at a time.
If I had only had the design weight, my guess would be that I should use
this weight as a pweight (that is, a probability weight) in combination
with cluster(country). But as my combined weight is not a "clean"
probability weight, can I do this? Or is there a better way to use
several weights and get robust standard errors for clustered data?
Swedish Institute for Social Research (SOFI)
SE-106 91 Stockholm
Phone: +46 8 674 79 97
* For searches and help try: