# Required functions: `strs.total`, `srs.total`

uspop=read.table("http://www.webpages.uidaho.edu/%7Erenaes/Data/R%20Data%20files/USpop.txt",sep='',header=T)
View(uspop)
set.seed(59373); library(sampling)

# and for some reason, there are errors in the dataset but now is a great time
# to illustrate how to remove and change values in the dataset from R (it will not change the original data file)
# because the first row is the US information for all states combined
uspop=uspop[-1,]
# Delaware's total population and Y1824 values are wrong
# 1. Find Delaware in the list
# 2. Change the corresponding elements of the variables
uspop[uspop$State=='Delaware',]
# The 2nd variable (Total=11821) and 4th variable (Y1824=11821)
# in fact, you can see that 11821 is the value for all the variables of this state
# We will change Total and Y1824 and just use those
uspop[uspop$State=='Delaware','Total']<-807385
uspop[uspop$State=='Delaware','Y1824']<-81501
# changes made
uspop[uspop$State=='Delaware',]
# Create new variable (strata) based on population
uspop$stratum<-ifelse(uspop$Total<3000000,1,ifelse(uspop$Total>10000000,3,2))
Ni=c(sum(uspop$stratum==1),sum(uspop$stratum==2),sum(uspop$stratum==3))
ni=rep(5,3)

uspop=uspop[order(uspop$stratum),]
strs=strata(uspop,stratanames=c("stratum"),size=ni, method="srswor")
uspop.strs=getdata(uspop,strs)


tauhat.strs('uspop.strs',uspop.strs$Y1824,uspop.strs$stratum,ni,Ni)

# compare to SRS

N=50; n=15
tauhat.srs('uspop.strs',uspop.srs$Pop18_24,n,N)