I am using svyset in stata with replicate weights using IPUMS ACS data. I know my svyset specifications are correct. I am trying to calculate accurate standard errors for an unemployment rate. When I calculate unemployment rate by dividing the sum of weights for all employed people by the sum of weights for the total labor force I get the exact same point estimate as I do when I create a dummy variable that is 1 for all unemployed observations and 0 for all others, drop all observations except for people in the labor force, and use
. My question is, will the standard error calculation be correct?
When using svy: means in this way I am not sure how variance in the estimate for the total labor force and for the total unemployed is being taken into account. But I'm also not really sure if this matters. The standard errors I get from this method match what I would expect to see.
One of those things where I am 95% sure there is no issue with this method, but not completely sure. I'm afraid I might be missing something.
Just in case anyone wants to see it my svyset code is:
Code:
svyset: mean dummy
When using svy: means in this way I am not sure how variance in the estimate for the total labor force and for the total unemployed is being taken into account. But I'm also not really sure if this matters. The standard errors I get from this method match what I would expect to see.
One of those things where I am 95% sure there is no issue with this method, but not completely sure. I'm afraid I might be missing something.
Just in case anyone wants to see it my svyset code is:
Code:
svyset [pw= perwt ], sdr( repwtp1 - repwtp80) vce(sdr) mse
