Re: st: error using nlcom and jacknife replicate weights

[jpitblado@stata.com (Jeff Pitblado, StataCorp LP)]

David W. Harless <dwharles@vcu.edu> is using -nlcom- after -svy jackknife:-
and is getting an error message when he shouldn't:

> I received an "internal error in adjustment for svy" using -nlcom- after
> -svy, regress- with jackknife replicate weights.  I replicated the error
> using the nhanes2jknife data set from page 63 of the Survey Data Manual (see
> below).
> 
> I've used -nlcom- without trouble with different survey weight structures.
> Is there some workaround or fix for the case of jackknife replicate weights?
> 
> I am using a fully updated Stata MP2 9.2 for Windows 64-bit.
> 
> ********************************************************************************
> . webuse nhanes2jknife
> 
> . svy, subpop(female): regress weight height age
  (output omitted)
> 
> . nlcom _b[height]/_b[age]
> 
>         _nl_1:  _b[height]/_b[age]
> internal error in adjustment for svy
> r(10001);

David found a bug in -nlcom- (actually in -testnl-, which -nlcom- uses as a
subroutine).  It's logic for computing the svy variance adjustement depends on
-e(N_strata)- and -e(N_psu)-, but -e(N_psu)- is not always defined when -svy
jackknife:- is employed.  -nlcom- should be using -e(df_r)-, which is always
defined by the -svy- estimation commands.

This will be fixed in the next ado-file update for Stata 9.

--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/