Solved – Post stratification weights and the survey package in R

rstratificationsurveysurvey-weights

I have survey data that needs to be weighted, and to help me with this task, I have access to the full joint distributions of the variables I want to use. As I understand it, I should use the postStratify() function in the survey package for this task, but I must admit that I am quite lost when it comes to the required syntax.

I have tried to use marginal distributions and the rake() function, this is quite straight-forward:
http://www.r-bloggers.com/survey-computing-your-own-post-stratification-weights-in-r/

But how do I incorporate all my joint distributions into the postStratify() function? The manual only lists a simple example and is a bit vague, to me at least. I have full joint distributions for 5 different variables.

Moreover, if want to use my weighted result with a function that does not take weight as an input, how should I go about and "duplicate" my rows in the final data? When I tried the rake() function I could only trim weights to an interval with a min value of 0.87. I was thinking I could round all weights to integers.

Best Answer

If you are having syntax trouble calculating post stratification weights and trimming I can recommend using www.spinnakerresearch.nl. No syntax needed. I believe you can trim max weights. However rounding weights to integers does not make sense as you will need to weight some cases down. Meaning a weight factor between 0 and 1.

Related Solutions

Solved – Post stratification weights in survey package in R

Ok Professor Lumley of the survey package answered my question on the R-help mailing list. Here is what he said:

There *is* only one weight for each subject.

You are misinterpreting the internal structures of the package: if you
want to see the weights, use the weights() function.

The components of $postStrata are used in standard error computations.

So that wraps up this question.

Solved – Using post-stratification weights in R survey package

If people say they have post-stratified weights, it does not necessarily mean they implemented post-stratification, proper (as in, rescaled the weights in each demographic cell to the known population total). About 80% of usage that I hear of "post-stratified weights" actually refers to calibrated weights (i.e., rather than trying to adjust each and every cell in a five-way table, the weights are only adjusted to match each of the five variables of the table individually). I produced what somebody referred to as a methodological rant on the distinction. The distinction, however, plays a role in standard error calculations, as Anthony noted in another answer. With properly post-stratified weights, you can apply the regular variance estimation formulae, more or less treating your post-strata as sampling strata (minor technicalities aside). With weights that are only calibrated on each table margin, computations are somewhat more involved. Both procedures are internalized in survey package, anyway, though. You just need to feed your post-stratification/calibration variables to the appropriate design object/formula.

library(survey)
data(api)
# cross-classified post-stratification variable in population
apipop$stype.sch.wide <- 10*as.integer(apipop$stype) +
as.integer(apipop$sch.wide)
# cross-classified post-stratification variable in sample
apiclus1$stype.sch.wide <- 
  10*as.integer(apiclus1$stype) + as.integer(apiclus1$sch.wide)
# population totals
(pop.totals <- xtabs(~stype.sch.wide, data=apipop))
# reference design
dclus1 <- svydesign(id=~dnum,weights=~pw,data=apiclus1,fpc=~fpc)
# post-stratification of the original design
dclus1p <- postStratify(dclus1,~stype.sch.wide, pop.totals)
# design with post-stratified weights, but no evidence of post-stratification
dclus1pfake <- svydesign(id=~dnum,weights=~weights(dclus1p),data=apiclus1,fpc=~fpc)
# taking off the design with known weights, add post-stratification interaction
dclus1pp <- postStratify(dclus1pfake,~stype.sch.wide, pop.totals)

# estimates and standard errors: starting point
svymean(~api00,dclus1)
# post-stratification reduces standard errors a bit
svymean(~api00,dclus1p)
# but here we are not aware of the survey being post-stratified
svymean(~api00,dclus1pfake)
# if we just add post-stratification variables to the design object
# that only had post-stratified weights, the result is the same
# as for post-stratified object based on the original weights
svymean(~api00,dclus1pp)

Related Question