Greetings,
This email is to announce a minor update/bug fix to my svr package, on SSC.
The -svrmodel- command calls the official Stata's svy command, in order to
estimate the variance under SRS, which is then used to calculate DEFF and
DEFT. In rare cases where the dataset was already svyset, and the strata,
psu, and/or FPC variables were missing for some of the cases used in the
SVR analysis, those cases were excluded from the calculation of the SRS
variance matrix, and therefore, from the calculations of DEFF/DEFT.
(This should not have come up often, since in most ordinary cases the
dataset will not be -svyset-, and even if it is, the replicate weight
variables will be missing for any cases that are missing strata, psu, or FPC.)
