Leslie R Hinkson (lhinkson@Princeton.EDU) needs to use the jackknife 2
variance replication method:
> I have a dataset that already has 62 replicate weights as well as a main
> weight that I thought was an iweight but am beginning to suspect may be a
> pweight. I was told that in order to do any analysis (including simply
> generating means and standard deviations) that I'd have to perform a
> jacknife 2 replication method. I tried to use the jknife command in Stata
> but couldn't get anywhere. Is this because I'm a qualitative researcher
> struggling with Stata or is it because I'm using the wrong command?
Leslie didn't mention Stata 8 or Stata 9. For Stata 8, Leslie will have to
use Nick Winter's -svr- suite of survey commands.
. ssc install svr
. help svr
Jackknife variance estimation using replication weight variables is a new
feature in Stata 9. Assuming Leslie's sampling weight variable is named
-sampwgt-, and the replicate weight variables are named -jkrw1-, -jkrw2-, ...,
-jkrw62-; the -svyset- command is as follows:
. svyset [pw=sampwgt], jkrweight(jkrw*, multiplier(1)) vce(jackknife)
(note: the shortcut 'jkrw*' will expand to 'jkrw1 jkrw2 ... jkrw62')
Then Leslie can compute means using -svy: mean-. Note that I specified the
suboption -multiplier(1)- in the -jkrweight()- option, it specifies the value
of the m_h term in the jackknife variance formula.
By default, -svy jackknife- assumes the 'jackknife n' replication
method, using the formula
m_h = (n_h-1)/n_h
for his multiplier (n_h is the stratum sample size), but is not
correct for the 'jackknife 2' replication method.
Leslie might also want to specify the -mse- option on -svyset- so that -svy
jackknife- uses the mean squared error formula for the variance calculation.
See [SVY] vce and [SVY] svyset for more details about jackknife variance
estimation and survey design settings.
--Jeff
jpitblado@stata.com
*
* 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/