Assuming that I need to use weights (perwt or hhwt depending on individual or household income estimates), how do I get those if I cannot use the “sum X, de” command with the SVY option? Thank you!

I have written epctile command that estimates percentiles for complex survey designs; findit epctile should point you to my website.

It is not STATA – it is not an acronym for anything. It is Stata.

Stas Kolenikov

Thanks- I cannot seem to download the package from stata- is there another link/do-file that I can manually run?

This is the prompt I get:

package name: epctile.pkg

from: http://fmwww.bc.edu/repec/bocode/e/

file http://fmwww.bc.edu/repec/bocode/e/ep… not found

could not load epctile.pkg from http://fmwww.bc.edu/repec/bocode/e/

It’s not on SSC, so your default thinking of ssc install won’t work. It’s on my personal webpage. I said that you need to use findit command that would lead to the right website.

Stas Kolenikov

I did use the findit command, and still no go. You could just paste the link to the response.

I was able to download the package. Can you help make sense of my output (percentiles for income)–What is the format of the results? e(b) etc?

. epctile inctot, p (5 10 25 50 75 90 95) svy valuemask (“00000”)

(running mean on estimation sample)

Survey: Mean estimation

Number of strata = 3342 Number of obs = 1037745

Number of PSUs = 619113 Population size = 19595440

Design df = 615771

Linearized

Mean Std. Err. [95% Conf. Interval]

__000006 -.0490513 . . .

__000007 -.0990513 . . .

__000008 -.0003342 . . .

__000009 -.0008501 . . .

__00000A -2.04e-06 . . .

__00000B -.0001335 . . .

__00000C -.0000524 . . .

Note: missing standard errors because of stratum with single

sampling unit.

estimates post: matrix has missing values