Bootstrapping using sample weights

[ALI SI]

Hi listers,

I want to use bootstrap with probability weights to get descriptive statistics. I have a sampling -weight- variable which came with data. Now I want to estimate summary statistics using bootstrap; however, but for the bootstrap process I want to sample observations based on sampling weight. I would appreciate if anyone could provide a code to do this in Stata.

Thanks
Ali

[Steve Samuels]

To do a bootstrap analysis, you must create a proper weight for each bootstap replicate. You do this with the command bsweights by Stas Kolenikov (type "findit bsweights"). There is an accompanying Stata Journal article with worked examples.

I haven't used bsweights myself, because the default survey linearization method estimates variances very well.

Reference:
Kolenikov, Stanislav. 2010. Resampling variance estimation for complex survey data. Stata Journal 10, no. 2: 165-199.

http://ageconsearch.umn.edu/bitstrea...art_st0187.pdf