[R] Using final sample weight in survey package

[Bert Gunter]

On Mon, Apr 4, 2016 at 9:35 AM, Anthony Damico <ajdamico at gmail.com> wrote:
> hi, probably not..  if your survey dataset has a complex design (like
> clusters/strata), you need to include them in the `svydesign` call.


> Coercing an incorrect survey design into a replicate-weighted design will
> not fix the problem of failing to account for the sampling strategy.

Amen, Anthony!

I would say **Fortune Nomination!** -- except that this may be a bit
too technical and statistical. But it's well said, nevertheless.

Thanks.

-- Bert



>
> On Mon, Apr 4, 2016 at 12:01 AM, José Fernando Zea <jfzeac at gmail.com> wrote:
>
>> I have the final sample weight (expansion factor) from a socieconomic
>> survey. I don't know the exact design used in the study ( (probably is a
>> stratified two-stage design).
>>
>> To illustrate my problem I will use the next dataset which have a sample
>> weight (but the design is not specified) and incorporate the design with
>> svydesign and create some bootstrap replicates in order to be able to
>> produce estimations.
>>
>> Is that correct?:
>>
>>
>> load(url("http://knutur.at/wsmt/R/RData/small.RData"))
>> library(survey)
>> small.w <- svydesign(ids = ~1, data = small, weights = small$weight)
>> design<-as.svrepdesign(small.w,type="bootstrap", replicates=100)
>>
>>
>>
>> Cordialmente
>> Jose F. Zea
>>
>>         [[alternative HTML version deleted]]
>>
>> ______________________________________________
>> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
>> https://stat.ethz.ch/mailman/listinfo/r-help
>> PLEASE do read the posting guide
>> http://www.R-project.org/posting-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.