[R] Significance of Svyrepdesign Object Warning
William Dunlap
wdunlap at tibco.com
Fri Oct 28 03:59:34 CEST 2016
For, now I would just use na.action=na.omit instead of na.exclude. My
comments were mainly for the package author.
Bill Dunlap
TIBCO Software
wdunlap tibco.com
On Thu, Oct 27, 2016 at 5:53 PM, Courtney Benjamin <cbenjami at btboces.org>
wrote:
> Hello Mr. Dunlap,
>
> I have gone back and re-read the responses to my question. I am
> interested in trying to apply your recommendation so I am doing things
> correctly; however I am not sure how to go about doing it within my code.
> It appears that you are digging quite deeply into R where I am not yet
> familiar. I am including a reproducible example; would you be willing to
> show an example of how it would be done? I greatly appreciate your
> advisement and time.
>
> Sincerely,
>
> Courtney
>
>
> library(RCurl)
> library(survey)
> data <- getURL("https://raw.githubusercontent.com/
> cbenjamin1821/careertech-ed/master/elsq1adj.csv")
> elsq1ch <- read.csv(text = data)
> #Specifying the svyrepdesign object which applies the BRR weights
> elsq1ch_brr<-svrepdesign(variables = elsq1ch[,1:16], repweights =
> elsq1ch[,18:217], weights = elsq1ch[,17], combined.weights = TRUE, type =
> "BRR")
> elsq1ch_brr
> #Logistic regression call which yields a warning regarding svyrepdesign
> object
> allCC <-svyglm(formula=F3ATTAINB~F1PARED+BYINCOME+F1RACE+F1SEX+
> F1RGPP2+F1HIMATH+F1RTRCC,family="binomial",design=
> elsq1ch_brr,subset=BYSCTRL==1&G10COHRT==1,na.action=na.exclude)
> summary(allCC)
>
>
>
>
>
> Courtney Benjamin
>
> Broome-Tioga BOCES
>
> Automotive Technology II Teacher
>
> Located at Gault Toyota
>
> Doctoral Candidate-Educational Theory & Practice
>
> State University of New York at Binghamton
>
> cbenjami at btboces.org
>
> 607-763-8633
> ------------------------------
> *From:* William Dunlap <wdunlap at tibco.com>
> *Sent:* Sunday, October 23, 2016 2:24 PM
> *To:* Anthony Damico
> *Cc:* Courtney Benjamin; r-help at r-project.org; Thomas Lumley
> *Subject:* Re: [R] Significance of Svyrepdesign Object Warning
>
> The immediate problem could be solved by changing the following lines in
> survey:::summary.svrepglm from
> presid <- resid(object, "pearson")
> dispersion <- sum(object$survey.design$pweights * presid^2,
> na.rm = TRUE)/sum(object$survey.design$pweights)
> to
> presid <- resid(object, "pearson")
> pweights <- naresid(object$na.action, object$survey.design$pweights)
> dispersion <- sum(pweights * presid^2, na.rm = TRUE)/sum(pweights,
> na.rm = TRUE)
>
> 'naresid' uses the information from na.exclude to match up the residuals
> with the row in the data that they correspond to. resid() calls it so it
> should
> also be applied to pweights so they line up correctly.
>
>
>
>
> Bill Dunlap
> TIBCO Software
> wdunlap tibco.com
>
> On Sun, Oct 23, 2016 at 11:17 AM, Anthony Damico <ajdamico at gmail.com>
> wrote:
>
>> hi, great example. i am ccing survey package author/maintainer dr.
>> lumley. why do you have `na.action=na.exclude`? if you remove it, things
>> work as expected--
>>
>>
>> library(RCurl)
>> library(survey)
>> data <- getURL("
>> https://raw.githubusercontent.com/cbenjamin1821/careertech-e
>> d/master/elsq1adj.csv
>> ")
>> elsq1ch <- read.csv(text = data)
>> #Specifying the svyrepdesign object which applies the BRR weights
>> elsq1ch_brr<-svrepdesign(variables = elsq1ch[,1:16], repweights =
>> elsq1ch[,18:217], weights = elsq1ch[,17], combined.weights = TRUE, type =
>> "BRR")
>> elsq1ch_brr
>> #Logistic regression call which yields a warning regarding
>> svyrepdesign
>> object
>>
>> # your warning
>> a <-
>> svyglm(formula=F3ATTAINB~F1PARED+BYINCOME+F1RACE+F1SEX+F1RGP
>> P2+F1HIMATH+F1RTRCC,family="binomial",design=elsq1ch_brr,
>> subset=BYSCTRL==1&G10COHRT==1,na.action=na.exclude)
>> summary(a)
>>
>> # works fine
>> a <-
>> svyglm(formula=F3ATTAINB~F1PARED+BYINCOME+F1RACE+F1SEX+F1RGP
>> P2+F1HIMATH+F1RTRCC,family="binomial",design=elsq1ch_brr,
>> subset=BYSCTRL==1&G10COHRT==1)
>> summary(a)
>>
>>
>>
>> the mismatch of vectors generating that warning happens inside
>>
>> debug(survey:::summary.svrepglm)
>>
>> [..snip..]
>>
>> Browse[2]> length(presid)
>> [1] 12614
>> Browse[2]> length(object$survey.design$pweights)
>> [1] 8397
>>
>>
>> and including vs excluding the na.action=na.exclude gives you a
>> slightly different dispersion parameter calculation
>>
>> (Dispersion parameter for binomial family taken to be 0.7756235)
>>
>> (Dispersion parameter for binomial family taken to be 0.7849244)
>>
>>
>> not sure if the two survey:::residuals.sv* methods should deal with the
>> na.action= parameter?
>>
>>
>> thanks
>>
>> On Sun, Oct 23, 2016 at 11:56 AM, Courtney Benjamin <cbenjami at btboces.org
>> >
>> wrote:
>>
>> > Hello R Users,
>> >
>> > I am using Lumley's Survey Package in R to analyze complex survey data
>> > that involves 200 balanced repeated replicate (BRR) weight variables. I
>> > have ensured that my svyrepdesign object that specifies the application
>> of
>> > the BRR weights to the data set is accurate and I have matched the
>> > published standard errors of the data set.
>> >
>> > When doing a logistic regression through the svyglm call, I receive the
>> > following warning:
>> >
>> > In object$survey.design$pweights * presid^2 :
>> > longer object length is not a multiple of shorter object length?
>> > I have search around quite a bit online and have not been able to find
>> any
>> > good interpretation of its meaning. I want to be sure that I am not
>> making
>> > some type of mistake that is causing this warning to be produced. Any
>> > advisement is greatly appreciated.
>> > The following is an MRE that can be pasted into the R console:
>> > library(RCurl)
>> > library(survey)
>> > data <- getURL("https://raw.githubusercontent.com/
>> > cbenjamin1821/careertech-ed/master/elsq1adj.csv")
>> > elsq1ch <- read.csv(text = data)
>> > #Specifying the svyrepdesign object which applies the BRR weights
>> > elsq1ch_brr<-svrepdesign(variables = elsq1ch[,1:16], repweights =
>> > elsq1ch[,18:217], weights = elsq1ch[,17], combined.weights = TRUE, type
>> =
>> > "BRR")
>> > elsq1ch_brr
>> > #Logistic regression call which yields a warning regarding svyrepdesign
>> > object
>> > svyglm(formula=F3ATTAINB~F1PARED+BYINCOME+F1RACE+F1SEX+
>> > F1RGPP2+F1HIMATH+F1RTRCC,family="binomial",design=
>> > elsq1ch_brr,subset=BYSCTRL==1&G10COHRT==1,na.action=na.exclude)
>> > allCC <- summary(svyglm(formula=F3ATTAINB~F1PARED+BYINCOME+
>> > F1RACE+F1SEX+F1RGPP2+F1HIMATH+F1RTRCC,family="binomial",
>> > design=elsq1ch_brr,subset=BYSCTRL==1&G10COHRT==1,na.action=na.exclude))
>> > allCC
>> >
>> > #Session Info
>> > #R version 3.3.1 (2016-06-21)
>> > #Platform: x86_64-w64-mingw32/x64 (64-bit)
>> > #Running under: Windows >= 8 x64 (build 9200)
>> >
>> > #locale:
>> > # [1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United
>> > States.1252
>> > #[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C
>> > #[5] LC_TIME=English_United States.1252
>> > #attached base packages:
>> > # [1] grid stats graphics grDevices utils datasets
>> > methods base
>> > #other attached packages:
>> > #[1] survey_3.31-2 survival_2.39-4 Matrix_1.2-6 RCurl_1.95-4.8
>> > bitops_1.0-6
>> > #loaded via a namespace (and not attached):
>> > #[1] tools_3.3.1 splines_3.3.1 knitr_1.14 lattice_0.20-33
>> >
>> >
>> > Courtney Benjamin
>> >
>> > Broome-Tioga BOCES
>> >
>> > Automotive Technology II Teacher
>> >
>> > Located at Gault Toyota
>> >
>> > Doctoral Candidate-Educational Theory & Practice
>> >
>> > State University of New York at Binghamton
>> >
>> > cbenjami at btboces.org<mailto:cbenjami at btboces.org>
>> >
>> > 607-763-8633
>> >
>> > [[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/posti
>> ng-guide.html
>> and provide commented, minimal, self-contained, reproducible code.
>>
>
>
[[alternative HTML version deleted]]
More information about the R-help
mailing list