Advances in Sampling Estimation - Sampling Spatial Units for Agricultural Surveys - page 248

Fig. 10.1 Histograms of the weight corrections using the GREG ( left ) and the Logit distance with
bounds (0, 100) ( right )
Figure 10.1 shows the histograms of the weight corrections using GREG and the
Logit distance with bounds (0,100).
> totpop < - c('(Intercept)'¼nrow(framepop), xc¼sum(framepop $ xc),
+yc¼sum(framepop $ yc), xc2¼sum(framepop $ xc2),yc2¼sum(framepop $ yc2))
> dsrsg < - calibrate(dsrs,~xc+yc+xc2+yc2, totpop)
> estg < - svytotal(~yobs, dsrsg, deff ¼ TRUE)
> estg
total SE DEff
yobs 91617.20 395.15 0.0397
> diff < - weights(dsrsg)/weights(dsrs)
> summary(diff)
Min. 1st Qu. Median Mean 3rd Qu. Max.
0.7523 0.9188 1.0380 1.0000 1.0920 1.1290
> set.seed(160964)
> ypps < - exp(yobs/10)
> probinc < - inclusionprobabilities(ypps,n)
> framepop < - cbind(framepop,probinc)
> pps < - UPtille(probinc)
> pps < - round(pps)
> framepps < - framepop[pps ¼¼1,]
> dpps < - svydesign(id¼~1,data¼framepps,
+ fpc¼~framepps $ probinc,pps¼"brewer")
> dppsg < - calibrate(dpps,~xc+yc+xc2+yc2, totpop)
> estg < - svytotal(~yobs, dppsg, deff¼TRUE)
> estg