[R] Problem in calculation of subpopulation mean in Survey package (Specify survey design with replicate weights)

Anthony Damico ajdamico at gmail.com
Sat Dec 24 12:00:47 CET 2016


hi, please revise your minimal reproducible example.  the objects `W` and
`bootstrap.results` do not exist.  thanks

On Fri, Dec 23, 2016 at 4:56 PM, Kristi Glover <kristi.glover at hotmail.com>
wrote:

> Hi R users,
>
> I got a problem when I was trying to calculate the population mean for
> different groups (classes) in using "svrepdesign", It worked when I used
> entire rows  but when I introduced by classes (groups or strata), it did
> not work out. I read several documents but I could not figure it out to the
> fix the problem.
>
> Actually, I do have about 500 data points (rows) in which there are
> several classes (groups). I developed the design using the svrepdesign, but
> when I introduced group in calculating mean, it says
>
> ""Error in qr.default(weights(design, "analysis"), tol = 1e-05) :
>   NA/NaN/Inf in foreign function call (arg 1)"
>
> I think I have to introduce groups in the design and I tried it different
> ways but it did not work it out. Would you mind to give some hints?I would
> be very grateful if you could give some hints.
>
> here I have attached a example (format of the data) and the code I used.
> Thanks for your help.
> have a great holidays.
>
> MG
> ___
>
> library(survey)
>
> dat<-structure(list(ID = 1:6, group = structure(c(8L, 7L, 7L, 8L,
> 7L, 7L), .Label = c("groupA", "groupB", "groupC", "groupD", "groupE",
> "groupF", "groupG", "groupH"), class = "factor"), ProbSelect = c(0.72,
> 0.62, 0.62, 0.72, 0.72, 0.62), density = c(28, 227, 65, 132,
> 13, 227), totalSampled = c(96L, 96L, 96L, 96L, 96L, 96L), pop = c(166L,
> 166L, 166L, 166L, 166L, 166L), wgt = c(2.42, 2.79, 2.79, 2.42,
> 2.42, 2.79)), .Names = c("ID", "group", "ProbSelect", "density",
> "totalSampled", "pop", "wgt"), row.names = c(NA, 6L), class = "data.frame")
>
> dat
>
> design_dat<-svrepdesign(data=dat, type="bootstrap", weights =
> I(1/dat$ProbSelect), repweights=W[,-1],scale=bootstrap.results$scale,
> rscale=bootstrap.results$rscales, combined.weights=TRUE)
>
> svymean(~ density,design=subset(design_dat, group=='groupH'),na.rm=T,
> digits=2)
>
> "Error in qr.default(weights(design, "analysis"), tol = 1e-05) :
>   NA/Na
>
>
>         [[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]]



More information about the R-help mailing list