Hello
I have three doubts about the use of the code ineqerr:
1) I am using the code to estimate inequality indexes and your
variances from a household survey with cluster sample in three stages
(city, section and holsehold) being two cities (PSUs) selected with
proportional probability to the size in each stratum and with occurrence
in the sample of self-representative city that are selected with
probability equal to 1. Which are the implications of to use only the
option psu and to leave of using the option psuwt in this code? In the
help of the code it is said that if no weighting variable is specified
for the first stage but the psu is specified, the bootstrap replicaton
is two stages of a simple random draw on the sample. What does this
mean?
2) the code doesn't have the option by. Do problems exist in using the
option if to do estimates of variances in sub-populations in the place
of the option by in replications commands for bootstrap as in the case
of the ineqerr? Does the preference for the option by only happen in
commands that use the Taylor's linearization method for variance
estimate as in the case of the command svymean?
3) What is the difference between the three methods of variance estimate
in the output (N), (P) and (BC)?
Thanks anyway
Henrique Neder
Universidade Federal de Uberlandia - Brazil
Este e-mail foi analisado pelo RAV AntiVirus no servidor doodle.dr.ufu.br da
Universidade Federal de Uberlāndia - Uberlāndia - MG - Brasil
*
* 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/