Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: brrweight(), jkrweight(), and vce() options of svyset


From   Sergiy Radyakin <serjradyakin@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: brrweight(), jkrweight(), and vce() options of svyset
Date   Wed, 12 May 2010 11:57:43 -0400

Dear Steve,

thank you for your prompt response. Could you please provide some more
information? You are writing "there is no inherent contradiction in
specifying both, because Stata can use only one at a time". If Stata can
use only one at a time, why specify both?

In my case, the user picks a dataset, a method for estimating variance, and
then the program runs the analysis on auto-pilot without any further user
intervention. In that case given the vce() option one of the two (or both) of
jkrweight() and brrweight() options becomes redundant. My goal is to minimize
the number of inputs that the user has to fill-in, and preclude the errors that
may arise if the user specifies an impossible combination of options for svyset.
All the commands in the analysis part assume that the svyset has already been
performed and will not change its settings in any way.

Consider the following options combinations which the users can specify
in the current Stata dialog:

1) -nothing-
2) vce(linearized)
3) vce(linearized) jkrweight(varlist)
4) vce(linearized) brrweight(varlist)
5) vce(linearized) jkrweight(varlist) brrweight(varlist)
6) vce(brr)
7) vce(brr) brrweight(varlist)
8) vce(brr) jkrweight(varlist)
9) vce(brr) brrweight(varlist) jkrweight(varlist)
10) vce(jackknife)
11) vce(jackknife) jkrweight(varlist)
12) vce(jackknife) brrweight(varlist)
13) vce(jackknife) jkrweight(varlist) brrweight(varlist)
14) jkrweight(varlist)
15) brrweight(varlist)
16) jkrweight(varlist) brrweight(varlist)
... forgot anything??.....

This is the list of 16 possible options combinations, while in fact there
are only 3 methods of variance estimation (linearized, brr, jackknife).

Consider using only two options: vce() and weights(). Then we have:
A) --nothing--
B) vce(linearized)
C) vce(linearized) weights(varlist)
D) vce(brr)
E) vce(brr) weights(varlist)
F) vce(jackknife)
G) vce(jackknife) weights(varlist)

Which one from the 1-16 list does not fit into some of the A-G?

This list is less than a half of the first one, there is less
meaningless combinations
(only D??) and it seems more clear what is going to happen after such settings.
The interpretation of the weights() option is jkrweight() if vce() is
jackknife and is
brrweight() if vce() is brr. There is some redundancy however: option
B is same as A.
C can be treated either as erroneous or the weights be ignored and C becomes
again A.

Consider getting rid of the vce() option all together. We have the following 4
combinations:

A) --nothing--
B) brrweight(varlist)
C) jkrweight(varlist)
D) brrweight(varlist) jkrweight(varlist).

Option (A) is equivalent of 1, 2, 3, 4, 5, 14 15 and 16 in the first list.
Option (B) is equivalent of 7 and 9
Option (C) is equivalent of 11 and 13
Option (D) is erroneous, because brrweight() and jkrweight() are
mutually exclusive
Other options from the first list are either erroneous combinations
that will not work,
or fit somewhere in the ABCD, with the exception of 10, since
jackknife works without
weights varlist, and Stata removes the option if its content is empty.
In that case the
user could specify option jackknife(varlist) or jackknife  <-- without
() if there is no varlist.


2Stas: thank you for your message. Is there any command that will use
"something from
jkrweights()" and "something from brrweights()" in one go? Also, you
write "for each
particular estimation command you run, you will have a choice of
whether you want to
specify the -vce(linearized)-, -vce(brr)- or -vce(jknife)-". Could you
please demonstrate this
with the -svy- commands? E.g. the svy mean. The regular -mean- command
that takes
the option -vce(jknife)- does not seem to be using the jkrweights()
set by the svy (and why
should it?).


2everyone: Let me rephrase the question: given only ONE svyset command
in the program,
(and possibly multiple svy estimation commands following it) does it
ever make sence to
allow for both brrweight() and jkrweight() options simultaneously in
this command. If so, can
you demonstrate such a case?

Thank you, Sergiy Radyakin






On Tue, May 11, 2010 at 10:41 PM, Steve Samuels <sjsamuels@gmail.com> wrote:
> and brr weights, Sergiy, but there is no inherent contradiction in
> specifying both, because Stata can use only one at a time. Your
> impressions (1) and (2) about minimum requirements are not correct.
> The -vce- option is not redundant; it is controlling.* If it is absent
> from both -svyset- and -svy: - prefix commands, Stata defaults to
> -vce(linearized)-. ..
>
> *The weights can also be invoked through the -svy brr:-- and -svy
> jackknife:- commands.
>
> Steve
>
> On Tue, May 11, 2010 at 8:44 PM, Sergiy Radyakin <serjradyakin@gmail.com> wrote:
>> Dear All,
>>
>> can somebody, please, explain why the brrweight(varlist) and jkrweight(varlist)
>> in svyset are NOT mutually exclusive? My understanding is that after I
>>
>> svyset ...whatever...,vce(brr) brrweight(varlist)
>>
>> specyfying jkrweight() afterwards in the same command does not affect
>> anything. However the standard Stata dialog for svyset (db svyset) permits
>> specifying both BRR and Jackknife weights and svyset command accepts
>> both without a problem.
>>
>> I guess the minimum information needed to be specified is:
>> 1) brrweight(varlist) -- translates to current vce(brr) brrweight(varlist)
>> 2) jkrweight(varlist) -- translates to current vce(jackknife) jkrweight(varlist)
>> 3) nothing -- translates to current vce(linearized)
>>
>> Meaning that the vce() option is redundant and brrweight() and jkrweight() must
>> be mutually exclusive.
>>
>> Does it harm in any way specifying both?
>> Is there benefit/convenience in specifying both?
>>
>> Any hints are very welcomed.
>>
>> Thank you, Sergiy Radyakin.
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/statalist/faq
>> *   http://www.ats.ucla.edu/stat/stata/
>>
>
>
>
> --
> Steven Samuels
> sjsamuels@gmail.com
> 18 Cantine's Island
> Saugerties NY 12477
> USA
> Voice: 845-246-0774
> Fax: 206-202-4783
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index