[R] Fw: Complex sampling survey _ Use of survey package

[Thomas Lumley]

On Fri, 12 Sep 2008, Ahoussou Sylvie wrote:

>
>
> --------------------------------------------------
> From: "Ahoussou Sylvie" <sylvie.ahoussou at antilles.inra.fr>
> Sent: Friday, September 12, 2008 9:48 AM
> To: "Thomas Lumley" <tlumley at u.washington.edu>
> Subject: Re: [R] Complex sampling survey _ Use of survey package
>
>> Thanks for your answer
>> 
>> I think I made a mistake when I recopied the 5 first rows of my database
>> 
>> here is the table with the comlums of interest
>> 
>> num esp fpc1 Totanim Id_An
>> 2045 G 551 12 10
>> 2046 C 551 68 11
>> 2070 G 551 9 50
>> 2070 S 551 9 51
>> 2070 S 551 9 52
>> 
>> yes Totanim is the total number of animals in the farm and num is the total 
>> number of herds

Do you mean 'fpc1 is the total number of herds'? That is what your 
svydesign() call says.

>> I keep on obtaining this error message
>> 
>> clustot<-svydesign(id=~num+ ~ Id_An, fpc=~fpc1+~Totanim, data=tab1)
>> 
>> Erreur dans as.fpc(fpc, strata, ids) :
>>  FPC implies >100% sampling in some strata.

Well, we seem to have either a bug or a problem with the data.

If you do
   options(error=recover)
before the svydesign() call you can go into as.fpc() and look at the data.

As an example;

Error in as.fpc(fpc, strata, ids) :
   FPC implies >100% sampling in some strata.

Enter a frame number, or 0 to exit

1: svydesign(id = ~dnum + snum, fpc = ~fpc1 + I(pmin(fpc2, 4)), data = 
apiclus2)
2: svydesign.default(id = ~dnum + snum, fpc = ~fpc1 + I(pmin(fpc2, 4)), 
data = apiclus2)
3: as.fpc(fpc, strata, ids)
Selection: 3
Called from: eval(expr, envir, enclos)
Browse[1]> which(sampsize>popsize, arr.ind=TRUE)
     row col
22   22   2
23   23   2
24   24   2
...

Browse[1]> sampsize[22,2]
[1] 5
Browse[1]> popsize[22,2]
[1] 4
Browse[1]> ids[22,]
    dnum    snum
22  200 200.841

So in this case one of the problems is in dnum 200, snum 841, where the 
population size was specified as 4 but the sample size is 5.

 	-thomas

Thomas Lumley			Assoc. Professor, Biostatistics
tlumley at u.washington.edu	University of Washington, Seattle