#Identifying pertinent variables for nonresponse survey: EVS 2008

#Load the necessary packages

library(foreign)
library(descr)
library(fmsb)
library ("Hmisc")
library(hexbin)
library("psych")
library("fmsb")
library("lawstat")
library(MASS)  
library("car")
library("gmodels")
library(survey)

#Read the correct file.
setwd("I://Work Caro/Documents/nonresponse surveys/EVS")
sink(file="EV08_1.txt",split=T)


EV8NR<-read.spss("EV08 for R.sav",use.missings=T,use.value.labels=T)
EV8<-as.data.frame(EV8NR)

#Data cleaning and variables recoding. Recoded variable finishing with M originate from main survey



EV8$f1_0[EV8$f1_0<0&EV8$f1_0>5]<-NA
EV8$v1[EV8$v1<0&EV8$v1>5]<-NA

EV8$f1_1[EV8$f1_1<0&EV8$f1_1>5]<-NA
EV8$v2[EV8$v2<0&EV8$v2>5]<-NA

EV8$f1_2[EV8$f1_2<0&EV8$f1_2>5]<-NA
EV8$v3[EV8$v3<0&EV8$v3>5]<-NA

EV8$f1_3[EV8$f1_3<0&EV8$f1_3>5]<-NA
EV8$v4[EV8$v4<0&EV8$v4>5]<-NA

EV8$f1_4[EV8$f1_4<0&EV8$f1_4>5]<-NA
EV8$v5[EV8$v5<0&EV8$v5>5]<-NA


EV8$f1_5[EV8$f1_5<0&EV8$vf1_5>5]<-NA
EV8$v6[EV8$v6<0&EV8$v6>5]<-NA

EV8$f2[EV8$f2<0&EV8$f2>5]<-NA
EV8$v63[EV8$v63<0&EV8$v63>5]<-NA

EV8$f3_0[EV8$f3_0<0&EV8$f3_0>5]<-NA
EV8$v275[EV8$v275<0&EV8$v275>5]<-NA

EV8$f3_0_r[EV8$f3_0_r<0&EV8$f3_0_r>5]<-NA
EV8$v2275_r[EV8$v2275_r<0&EV8$v2275_r>5]<-NA

EV8$f3_1[EV8$f3_1<0&EV8$f3_1>5]<-NA
EV8$v282[EV8$v282<0&EV8$v282>5]<-NA

EV8$f3_2[EV8$f3_2<0&EV8$f3_2>5]<-NA
EV8$v283[EV8$v283<0&EV8$v283>5]<-NA


EV8$group<-NA
EV8$group[EV8$f4_0==1|EV8$f4_1==1|EV8$f4_2==1|EV8$f4_3==1|EV8$f4_4==1|EV8$f4_5==1|EV8$f4_6==1|EV8$f4_7==1|EV8$f4_8==1|EV8$f4_9==1|EV8$f4_10==1|EV8$f4_11==1|
             EV8$f4_12==1|EV8$f4_13==1|EV8$f4_14==1]<-1
EV8$group[EV8$f4_0==2&EV8$f4_1==2&EV8$f4_2==2&EV8$f4_3==2&EV8$f4_4==2&EV8$f4_5==2&EV8$f4_6==2&EV8$f4_7==2&EV8$f4_8==2&EV8$f4_9==2&EV8$f4_10==2&EV8$f4_11==2&
             EV8$f4_12==2&EV8$f4_13==2&EV8$f4_14==2]<-0
table(EV8$group)

EV8$groupM<-NA
EV8$groupM[EV8$v10==1|EV8$v11==1|EV8$v12==1|EV8$v13==1|EV8$v14==1|EV8$v15==1|EV8$v16==1|EV8$v17==1|EV8$v18==1|EV8$v19==1|EV8$v20==1|EV8$v21==1|
               EV8$v22==1|EV8$v23==1|EV8$v24==1]<-1
EV8$groupM[EV8$v10==2&EV8$v11==2&EV8$v12==2&EV8$v13==2&EV8$v14==2&EV8$v15==2&EV8$v16==2&EV8$v17==2&EV8$v18==2&EV8$v19==2&EV8$v20==2&EV8$v21==2&
  EV8$v22==2&EV8$v23==2&EV8$v24==2]<-0
table(EV8$groupM)

EV8$f5[EV8$f5<0&EV8$f5>5]<-NA
EV8$v186[EV8$v186<0&EV8$v186>5]<-NA


EV8$ager<-NA
EV8$ager<-108-EV8$f7
EV8$ager[EV8$f7<15&EV8$f7>99]<-NA
table(EV8$ager)

EV8$agerM<-NA
EV8$agerM<-108-EV8$v303
EV8$agerM[EV8$v303<15&EV8$v303>99]<-0
table(EV8$agerM)

EV8$gndr<-NA
EV8$gndr[EV8$f8==2]<-0
EV8$gndr[EV8$f8==1]<-1
table(EV8$gndr)

EV8$gndrM<-NA
EV8$gndrM[EV8$v302==2]<-0
EV8$gndrM[EV8$v302==1]<-1
table(EV8$gndrM)


EV8$edu<-NA
EV8$edu[EV8$f9tp_0==1|EV8$f9tp_1==1|EV8$f9tp_2==1|EV8$f9tp_3==1|EV8$f9tp_12==1|EV8$f9tp_4==1|EV8$f9tp_5==1|EV8$f9tp_6==1|EV8$f9tp_7==1|EV8$f9tp_8==1|
          EV8$f9tp_9==1|EV8$f9tp_10==1|EV8$f9tp_11==1]<-2
EV8$edu[EV8$f9tp_0==2|EV8$f9tp_1==2|EV8$f9tp_2==2|EV8$f9tp_3==2|EV8$f9tp_12==2|EV8$f9tp_4==2|EV8$f9tp_5==2|EV8$f9tp_6==2|EV8$f9tp_7==2|EV8$f9tp_8==2
        |EV8$f9tp_9==2|EV8$f9tp_10==2|EV8$f9tp_11==2]<-3
EV8$edu[EV8$f9tp_0==3|EV8$f9tp_1==3|EV8$f9tp_2==3|EV8$f9tp_3==3|EV8$f9tp_12==3|EV8$f9tp_4==3|EV8$f9tp_5==3|EV8$f9tp_6==3|EV8$f9tp_7==3|EV8$f9tp_8==3
        |EV8$f9tp_9==3|EV8$f9tp_10==3|EV8$f9tp_11==3]<-6
EV8$edu[EV8$f9tp_0==4|EV8$f9tp_1==4|EV8$f9tp_2==4|EV8$f9tp_3==4|EV8$f9tp_12==4|EV8$f9tp_4==4|EV8$f9tp_5==4|EV8$f9tp_6==4|EV8$f9tp_7==4|EV8$f9tp_8==4
        |EV8$f9tp_9==4|EV8$f9tp_10==4|EV8$f9tp_11==4]<-4
EV8$edu[EV8$f9tp_0==5|EV8$f9tp_1==5|EV8$f9tp_2==5|EV8$f9tp_3==5|EV8$f9tp_12==5|EV8$f9tp_4==5|EV8$f9tp_5==5|EV8$f9tp_6==5|EV8$f9tp_7==5|EV8$f9tp_8==5
        |EV8$f9tp_9==5|EV8$f9tp_10==5|EV8$f9tp_11==5]<-5
EV8$edu[EV8$f9tp_0==6|EV8$f9tp_1==6|EV8$f9tp_2==6|EV8$f9tp_3==6|EV8$f9tp_12==6|EV8$f9tp_4==6|EV8$f9tp_5==6|EV8$f9tp_6==6|EV8$f9tp_7==6|EV8$f9tp_8==6
        |EV8$f9tp_9==6|EV8$f9tp_10==6|EV8$f9tp_11==6]<-9
EV8$edu[EV8$f9tp_0==7|EV8$f9tp_1==7|EV8$f9tp_2==7|EV8$f9tp_3==7|EV8$f9tp_12==7|EV8$f9tp_4==7|EV8$f9tp_5==7|EV8$f9tp_6==7|EV8$f9tp_7==7|EV8$f9tp_8==7
        |EV8$f9tp_9==7|EV8$f9tp_10==7|EV8$f9tp_11==7]<-12
EV8$edu[EV8$f9tp_0==8|EV8$f9tp_1==8|EV8$f9tp_2==8|EV8$f9tp_3==8|EV8$f9tp_12==8|EV8$f9tp_4==8|EV8$f9tp_5==8|EV8$f9tp_6==8|EV8$f9tp_7==8|EV8$f9tp_8==8
        |EV8$f9tp_9==8|EV8$f9tp_10==8|EV8$f9tp_11==8]<-7
EV8$edu[EV8$f9tp_0==9|EV8$f9tp_1==9|EV8$f9tp_2==9|EV8$f9tp_3==9|EV8$f9tp_12==9|EV8$f9tp_4==9|EV8$f9tp_5==9|EV8$f9tp_6==9|EV8$f9tp_7==9|EV8$f9tp_8==9
        |EV8$f9tp_9==9|EV8$f9tp_10==9|EV8$f9tp_11==9]<-8
EV8$edu[EV8$f9tp_0==10|EV8$f9tp_1==10|EV8$f9tp_2==10|EV8$f9tp_3==10|EV8$f9tp_12==10|EV8$f9tp_4==10|EV8$f9tp_5==10|EV8$f9tp_6==10|EV8$f9tp_7==10|EV8$f9tp_8==10
        |EV8$f9tp_9==10|EV8$f9tp_10==10|EV8$f9tp_11==10]<-10
EV8$edu[EV8$f9tp_0==11|EV8$f9tp_1==11|EV8$f9tp_2==11|EV8$f9tp_3==11|EV8$f9tp_12==11|EV8$f9tp_4==11|EV8$f9tp_5==11|EV8$f9tp_6==11|EV8$f9tp_7==11|EV8$f9tp_8==11
        |EV8$f9tp_9==11|EV8$f9tp_10==11|EV8$f9tp_11==11]<-11
EV8$edu[EV8$f9tp_0==13|EV8$f9tp_1==12|EV8$f9tp_2==12|EV8$f9tp_3==12|EV8$f9tp_12==12|EV8$f9tp_4==12|EV8$f9tp_5==12|EV8$f9tp_6==12|EV8$f9tp_7==12|EV8$f9tp_8==12
        |EV8$f9tp_9==12|EV8$f9tp_10==12|EV8$f9tp_11==12]<-13
EV8$edu[EV8$f9tp_0==13|EV8$f9tp_1==13|EV8$f9tp_2==13|EV8$f9tp_3==13|EV8$f9tp_12==13|EV8$f9tp_4==13|EV8$f9tp_5==13|EV8$f9tp_6==13|EV8$f9tp_7==13|EV8$f9tp_8==13
        |EV8$f9tp_9==13|EV8$f9tp_10==13|EV8$f9tp_11==13]<-14
EV8$edu[EV8$f9tp_0==14|EV8$f9tp_1==14|EV8$f9tp_2==14|EV8$f9tp_3==14|EV8$f9tp_12==14|EV8$f9tp_4==14|EV8$f9tp_5==14|EV8$f9tp_6==14|EV8$f9tp_7==14|EV8$f9tp_8==14
        |EV8$f9tp_9==14|EV8$f9tp_10==14|EV8$f9tp_11==14]<-15
EV8$edu[EV8$f9tp_0==15|EV8$f9tp_1==15|EV8$f9tp_2==15|EV8$f9tp_3==15|EV8$f9tp_12==15|EV8$f9tp_4==15|EV8$f9tp_5==15|EV8$f9tp_6==15|EV8$f9tp_7==15|EV8$f9tp_8==15
        |EV8$f9tp_9==15|EV8$f9tp_10==15|EV8$f9tp_11==15]<-16
EV8$edu[EV8$f9tp_0==16|EV8$f9tp_1==16|EV8$f9tp_2==16|EV8$f9tp_3==16|EV8$f9tp_12==16|EV8$f9tp_4==16|EV8$f9tp_5==16|EV8$f9tp_6==16|EV8$f9tp_7==16|EV8$f9tp_8==16
        |EV8$f9tp_9==16|EV8$f9tp_10==16|EV8$f9tp_11==16]<-17
table(EV8$edu)

EV8$eduM<-EV8$v336
EV8$eduM[EV8$v336<1&EV8$v336>=18]<-NA

table(EV8$eduM)

EV8$work<-NA
EV8$work[(EV8$f10tp_0>=1&EV8$f10tp_0<=10)|(EV8$f10tp_1>=1&EV8$f10tp_1<=10)|(EV8$f10tp_2>=1&EV8$f9tp_2<=10)|
  (EV8$f10tp_3>=1&EV8$f10tp_3<=10)|(EV8$f10tp_4>=1&EV8$f10tp_4<=10)]<-0
EV8$work[(EV8$f10tp_0>=1&EV8$f10tp_0<=4)|(EV8$f10tp_1>=1&EV8$f10tp_1<=4)|(EV8$f10tp_2>=1&EV8$f9tp_2<=4)|
  (EV8$f10tp_3>=1&EV8$f10tp_3<=4)|(EV8$f10tp_4>=1&EV8$f10tp_4<=4)]<-1
table(EV8$work)

EV8$workM<-NA
EV8$workM[EV8$v337>=1&EV8$v337<5]<-1
EV8$workM[EV8$v337>=5&EV8$v337<=10]<-0
table(EV8$workM)

EV8$alone<-NA
EV8$alone[EV8$f11==1]<-1
EV8$alone[EV8$f11>1]<-0
table(EV8$alone)


  
EV8$aloneM<-NA
EV8$aloneM[EV8$v323==1|EV8$v324a==1|EV8$v325a==1|EV8$v326a==1]<-0
EV8$aloneM[EV8$v323==2&EV8$v324a==2&EV8$v325a==2&EV8$v326a==2]<-1
table(EV8$aloneM)


#select the cases for which the respondent answered both survey and sex and year of birth corresponds.

nrandmain <- which(EV8$nrsurv==1&EV8$mainsurv==1)
EV81<-EV8[nrandmain,]
table(EV81$f8)
table(EV81$v302)
EV81$f8b<-0
EV81$f8b[EV81$f8==1]<-1
EV81$f8b[EV81$f8==2]<-2
EV81$test<-EV81$ager-EV81$agerM
EV81$agetest<-0
EV81$agetest[EV81$test<=-3|EV81$test>=3]<-1
nrandmain1<-which((EV81$f8==EV81$v302| EV81$f8 == 3|EV81$f8b==0)&EV81$agetest==0)
EV8corr<-EV81[nrandmain1,]


type<- c("corr","diff")
EV8corrmean<-as.data.frame(type)

EV8corrmean$polintr[EV8corrmean$type=="corr"]<-cor(EV8corr$f5,EV8corr$v186,"pairwise.complete.obs", method="spearman")
EV8corrmean$impwork[EV8corrmean$type=="corr"]<-cor(EV8corr$f1_0,EV8corr$v1,"pairwise.complete.obs", method="spearman")
EV8corrmean$impfami[EV8corrmean$type=="corr"]<-cor(EV8corr$f1_1,EV8corr$v2,"pairwise.complete.obs", method="spearman")
EV8corrmean$impfrie[EV8corrmean$type=="corr"]<-cor(EV8corr$f1_2,EV8corr$v3,"pairwise.complete.obs", method="spearman")
EV8corrmean$impleas[EV8corrmean$type=="corr"]<-cor(EV8corr$f1_3,EV8corr$v4,"pairwise.complete.obs", method="spearman")
EV8corrmean$imppol[EV8corrmean$type=="corr"]<-cor(EV8corr$f1_4,EV8corr$v5,"pairwise.complete.obs", method="spearman")
EV8corrmean$imprel[EV8corrmean$type=="corr"]<-cor(EV8corr$f1_5,EV8corr$v6,"pairwise.complete.obs", method="spearman")
EV8corrmean$trust[EV8corrmean$type=="corr"]<-cor(EV8corr$f2,EV8corr$v63,"pairwise.complete.obs", method="spearman")
EV8corrmean$immig[EV8corrmean$type=="corr"]<-cor(EV8corr$f3_0,EV8corr$v275,"pairwise.complete.obs", method="spearman")
EV8corrmean$just[EV8corrmean$type=="corr"]<-cor(EV8corr$f3_1,EV8corr$v282,"pairwise.complete.obs", method="spearman")
EV8corrmean$stick[EV8corrmean$type=="corr"]<-cor(EV8corr$f3_2,EV8corr$v283,"pairwise.complete.obs", method="spearman")
EV8corrmean$group[EV8corrmean$type=="corr"]<-phi(table(EV8corr$group,EV8corr$groupM))
EV8corrmean$alone[EV8corrmean$type=="corr"]<-phi(table(EV8corr$alone,EV8corr$aloneM))
EV8corrmean$ager[EV8corrmean$type=="corr"]<-cor(EV8corr$ager,EV8corr$agerM,"pairwise.complete.obs", method="pearson")
EV8corrmean$edu[EV8corrmean$type=="corr"]<-cor(EV8corr$edu,EV8corr$eduM,"pairwise.complete.obs", method="spearman")
EV8corrmean$work[EV8corrmean$type=="corr"]<-phi(table(EV8corr$work,EV8corr$workM))
                                                         
#create difference varaibles for mean
EV8corr$polintrD<-NA
EV8corr$polintrD<-EV8corr$f5-EV8corr$v186
EV8corrmean$polintr[EV8corrmean$type=="diff"]<-mean(EV8corr$polintrD, na.rm=TRUE)

EV8corr$impworkD<-NA
EV8corr$impworkD<-EV8corr$f1_0-EV8corr$v1
EV8corrmean$impwork[EV8corrmean$type=="diff"]<-mean(EV8corr$impworkD, na.rm=TRUE)

EV8corr$impfamiD<-NA
EV8corr$impfamiD<-EV8corr$f1_1-EV8corr$v2
EV8corrmean$impfami[EV8corrmean$type=="diff"]<-mean(EV8corr$impfamiD, na.rm=TRUE)

EV8corr$impfrieD<-NA
EV8corr$impfrieD<-EV8corr$f1_2-EV8corr$v3
EV8corrmean$impfrie[EV8corrmean$type=="diff"]<-mean(EV8corr$impfrieD, na.rm=TRUE)


EV8corr$impleasD<-NA
EV8corr$impleasD<-EV8corr$f1_3-EV8corr$v4
EV8corrmean$impleas[EV8corrmean$type=="diff"]<-mean(EV8corr$impleasD, na.rm=TRUE)

EV8corr$imppolD<-NA
EV8corr$imppolD<-EV8corr$f1_4-EV8corr$v5
EV8corrmean$imppol[EV8corrmean$type=="diff"]<-mean(EV8corr$imppolD, na.rm=TRUE)

EV8corr$imprelD<-NA
EV8corr$imprelD<-EV8corr$f1_5-EV8corr$v6
EV8corrmean$imprel[EV8corrmean$type=="diff"]<-mean(EV8corr$imprelD, na.rm=TRUE)

EV8corr$trustD<-NA
EV8corr$trustD<-EV8corr$f2-EV8corr$v63
EV8corrmean$trust[EV8corrmean$type=="diff"]<-mean(EV8corr$trustD, na.rm=TRUE)

EV8corr$immigD<-NA
EV8corr$immigD<-EV8corr$f3_0-EV8corr$v275
EV8corrmean$immig[EV8corrmean$type=="diff"]<-mean(EV8corr$immigD, na.rm=TRUE)
median(EV8corr$immigD, na.rm=TRUE)

EV8corr$justD<-NA
EV8corr$justD<-EV8corr$f3_1-EV8corr$v282
EV8corrmean$just[EV8corrmean$type=="diff"]<-mean(EV8corr$justD, na.rm=TRUE)

EV8corr$stickD<-NA
EV8corr$stickD<-EV8corr$f3_2-EV8corr$v283
EV8corrmean$stick[EV8corrmean$type=="diff"]<-mean(EV8corr$stickD, na.rm=TRUE)


EV8corr$groupD<-NA
EV8corr$groupD<-EV8corr$group-EV8corr$groupM
EV8corrmean$group[EV8corrmean$type=="diff"]<-mean(EV8corr$groupD, na.rm=TRUE)

EV8corr$agerD<-NA
EV8corr$agerD<-EV8corr$ager-EV8corr$agerM
EV8corrmean$ager[EV8corrmean$type=="diff"]<-mean(EV8corr$agerD, na.rm=TRUE)

EV8corr$gndrD<-NA
EV8corr$gndrD<-EV8corr$gndr-EV8corr$gndrM
EV8corrmean$gndr[EV8corrmean$type=="diff"]<-mean(EV8corr$gndrD, na.rm=TRUE)

EV8corr$eduD<-NA
EV8corr$eduD<-EV8corr$edu-EV8corr$eduM
EV8corrmean$edu[EV8corrmean$type=="diff"]<-mean(EV8corr$eduD, na.rm=TRUE)

EV8corr$workD<-NA
EV8corr$workD<-EV8corr$work-EV8corr$workM
EV8corrmean$work[EV8corrmean$type=="diff"]<-mean(EV8corr$workD, na.rm=TRUE)

EV8corr$aloneD<-NA
EV8corr$aloneD<-EV8corr$alone-EV8corr$aloneM
EV8corrmean$alone[EV8corrmean$type=="diff"]<-mean(EV8corr$aloneD, na.rm=TRUE)

#symmetry test.

symmetry.test(EV8corr$polintrD)
symmetry.test(EV8corr$impfamiD)
symmetry.test(EV8corr$impleasD)
symmetry.test(EV8corr$imppolD)
symmetry.test(EV8corr$imprelD)
symmetry.test(EV8corr$immigD)
symmetry.test(EV8corr$eduD)

#wilcoxon test.



wilcox.test(EV8corr$f5, EV8corr$v186,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f1_0, EV8corr$v1,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f1_1, EV8corr$v2,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f1_2, EV8corr$v3,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f1_3, EV8corr$v4,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f1_4, EV8corr$v5,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f1_5, EV8corr$v6,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f2, EV8corr$v63,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f3_0, EV8corr$v275,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f3_1, EV8corr$v282,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$f3_2, EV8corr$v283,paired=TRUE,na.rm=TRUE)

t.test(EV8corr$ager, EV8corr$agerM,paired=TRUE,na.rm=TRUE)
wilcox.test(EV8corr$edu, EV8corr$eduM,paired=TRUE,na.rm=TRUE)


prop.test(matrix(c(45,43,101,102),nrow=2,ncol=2))
table(EV8corr$work)
table(EV8corr$workM)
prop.test(matrix(c(110,104,37,22),nrow=2,ncol=2))
table(EV8corr$alone)
table(EV8corr$aloneM)

EV8$mnsurv<-0
EV8$mnsurv[EV8$mainsurv==1]<-1

#creation of harmonized variable: takes the values of the main survey for respondents and of NRS for nonrespondents.
EV8<-EV8[order(-EV8$mnsurv,-EV8$nrsurv),]

v1<-EV8$gndrM[EV8$mnsurv>=1]
v2<-EV8$gndr[EV8$mnsurv==0]
EV8$gndr_h<-c(v1,v2)
table(EV8$gndr_h)

v1<-EV8$v1[EV8$mnsurv==1]
v2<-EV8$f1_0[EV8$mnsurv==0] 
EV8$impwork_h<-c(v1,v2)
table(EV8$impwork_h)

v1<-EV8$v2[EV8$mnsurv==1]
v2<-EV8$f1_1[EV8$mnsurv==0]
EV8$impfami_h<-c(v1,v2)
table(EV8$impfami_h)

v1<-EV8$v3[EV8$mnsurv==1]
v2<-EV8$f1_2[EV8$mnsurv==0]
EV8$impfrie_h<-c(v1,v2)
table(EV8$impfrie_h)

v1<-EV8$v4[EV8$mnsurv==1]
v2<-EV8$f1_3[EV8$mnsurv==0]
EV8$impleas_h<-c(v1,v2)
table(EV8$impleas_h)

v1<-EV8$v5[EV8$mnsurv==1]
v2<-EV8$f1_4[EV8$mnsurv==0]
EV8$imppol_h<-c(v1,v2)
table(EV8$imppol_h)

v1<-EV8$v6[EV8$mnsurv==1]
v2<-EV8$f1_5[EV8$mnsurv==0]
EV8$imprel_h<-c(v1,v2)
table(EV8$imprel_h)

v1<-EV8$v63[EV8$mnsurv==1]
v2<-EV8$f2[EV8$mnsurv==0]
EV8$trust_h<-c(v1,v2)
table(EV8$trust_h)

v1<-EV8$v275[EV8$mnsurv==1]
v2<-EV8$f3_0[EV8$mnsurv==0]
EV8$immig_h<-c(v1,v2)
table(EV8$immig_h)

v1<-EV8$v282[EV8$mnsurv==1]
v2<-EV8$f3_1[EV8$mnsurv==0]
EV8$just_h<-c(v1,v2)
table(EV8$just_h)

v1<-EV8$v283[EV8$mnsurv==1]
v2<-EV8$f3_2[EV8$mnsurv==0]
EV8$stick_h<-c(v1,v2)
table(EV8$stick_h)

v1<-EV8$v186[EV8$mnsurv==1]
v2<-EV8$v186[EV8$mnsurv==0]
EV8$polintr_h<-c(v1,v2)
table(EV8$polintr_h)

v1<-EV8$agerM[EV8$mnsurv==1]
v2<-EV8$ager[EV8$mnsurv==0]
EV8$ager_h<-c(v1,v2)
table(EV8$ager_h)

v1<-EV8$gndrM[EV8$mnsurv==1]
v2<-EV8$gndr[EV8$mnsurv==0]
EV8$gndr_h<-c(v1,v2)
table(EV8$gndr_h)

v1<-EV8$eduM[EV8$mnsurv==1]
v2<-EV8$edu[EV8$mnsurv==0]
EV8$edu_h<-c(v1,v2)
table(EV8$edu_h)

v1<-EV8$workM[EV8$mnsurv==1]
v2<-EV8$work[EV8$mnsurv==0]
EV8$work_h<-c(v1,v2)
table(EV8$work_h)

v1<-EV8$aloneM[EV8$mnsurv==1]
v2<-EV8$alone[EV8$mnsurv==0]
EV8$alone_h<-c(v1,v2)
table(EV8$alone_h)



#compare the distrubutions for respondents and non-respondents.

crosstab(EV8$gndr_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$ager_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$edu_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$work_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$alone_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$impwork_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$impfami_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$impfrie_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$impleas_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$imppol_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$imprel_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$trust_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$immig_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$just_h,EV8$mnsurv,prop.c=T,chisq=T)
crosstab(EV8$stick_h,EV8$mnsurv,prop.c=T,chisq=T)

#review
evsNR<-EV8[which(EV8$nrsurv==1),]
evsNR<-evsNR[which(!is.na(evsNR$ager)&!is.na(evsNR$gndr)&!is.na(evsNR$edu)&!is.na(evsNR$work)&!is.na(evsNR$alone)&!is.na(evsNR$f1_0)
                   &!is.na(evsNR$f1_1)&!is.na(evsNR$f1_2)&!is.na(evsNR$f1_3)&!is.na(evsNR$f1_5)&!is.na(evsNR$f1_4)&!is.na(evsNR$f2)
                   &!is.na(evsNR$f3_0)&!is.na(evsNR$f3_1)&!is.na(evsNR$f3_2)&!is.na(evsNR$group)
                   &!is.na(evsNR$f5)),]
evsNR$NR<-1
evsNR$NR[evsNR$mainsurv==1]<-0
table(evsNR$NR)
evsreview<-glm(NR~ager+gndr+edu+work+alone+f1_0+f1_1+f1_2+f1_3+f1_4+f1_5+f2+f3_0+f3_1+f3_2+group+f5,data=evsNR,family=binomial)
summary(evsreview)
NagelkerkeR2(evsreview)

evsNR$propscore<-predict(evsreview, type="response")
evsNR$propweight<-1/evsNR$propscore

type<- c("RUmean","RWmean", "NRmean")
evsr<-as.data.frame(type)
evsNRNR<-evsNR[which(evsNR$NR==1),]

evsr$impwork[evsr$type=="RUmean"]<-mean(evsNR$f1_0, na.rm=TRUE)
evsr$impwork[evsr$type=="RWmean"]<-weighted.mean(evsNR$f1_0,evsNR$propweight, na.rm=TRUE)
evsr$impwork[evsr$type=="NRmean"]<-mean(evsNR$v1, na.rm=TRUE)

evsr$impfam[evsr$type=="RUmean"]<-mean(evsNR$f1_1, na.rm=TRUE)
evsr$impfam[evsr$type=="RWmean"]<-weighted.mean(evsNR$f1_1,evsNR$propweight, na.rm=TRUE)
evsr$impfam[evsr$type=="NRmean"]<-mean(evsNR$v2, na.rm=TRUE)

evsr$imfrie[evsr$type=="RUmean"]<-mean(evsNR$f1_2, na.rm=TRUE)
evsr$imfrie[evsr$type=="RWmean"]<-weighted.mean(evsNR$f1_2,evsNR$propweight, na.rm=TRUE)
evsr$imfrie[evsr$type=="NRmean"]<-mean(evsNR$v3, na.rm=TRUE)

evsr$implea[evsr$type=="RUmean"]<-mean(evsNR$f1_3, na.rm=TRUE)
evsr$implea[evsr$type=="RWmean"]<-weighted.mean(evsNR$f1_3,evsNR$propweight, na.rm=TRUE)
evsr$implea[evsr$type=="NRmean"]<-mean(evsNR$v4, na.rm=TRUE)

evsr$imppol[evsr$type=="RUmean"]<-mean(evsNR$f1_4, na.rm=TRUE)
evsr$imppol[evsr$type=="RWmean"]<-weighted.mean(evsNR$f1_4,evsNR$propweight, na.rm=TRUE)
evsr$imppol[evsr$type=="NRmean"]<-mean(evsNR$v5, na.rm=TRUE)

evsr$imprel[evsr$type=="RUmean"]<-mean(evsNR$f1_5, na.rm=TRUE)
evsr$imprel[evsr$type=="RWmean"]<-weighted.mean(evsNR$f1_5,evsNR$propweight, na.rm=TRUE)
evsr$imprel[evsr$type=="NRmean"]<-mean(evsNR$v6, na.rm=TRUE)

evsr$trust[evsr$type=="RUmean"]<-mean(evsNR$f2, na.rm=TRUE)
evsr$trust[evsr$type=="RWmean"]<-weighted.mean(evsNR$f2,evsNR$propweight, na.rm=TRUE)
evsr$trust[evsr$type=="NRmean"]<-mean(evsNR$v63, na.rm=TRUE)

evsr$immig[evsr$type=="RUmean"]<-mean(evsNR$f3_0, na.rm=TRUE)
evsr$immig[evsr$type=="RWmean"]<-weighted.mean(evsNR$f3_0,evsNR$propweight, na.rm=TRUE)
evsr$immig[evsr$type=="NRmean"]<-mean(evsNR$v275, na.rm=TRUE)


evsr$just[evsr$type=="RUmean"]<-mean(evsNR$f3_1, na.rm=TRUE)
evsr$just[evsr$type=="RWmean"]<-weighted.mean(evsNR$f3_1,evsNR$propweight, na.rm=TRUE)
evsr$just[evsr$type=="NRmean"]<-mean(evsNR$v282, na.rm=TRUE)

evsr$stick[evsr$type=="RUmean"]<-mean(evsNR$f3_2, na.rm=TRUE)
evsr$stick[evsr$type=="RWmean"]<-weighted.mean(evsNR$f3_2,evsNR$propweight, na.rm=TRUE)
evsr$stick[evsr$type=="NRmean"]<-mean(evsNR$v283, na.rm=TRUE)

evsr$just[evsr$type=="RUmean"]<-mean(evsNR$group, na.rm=TRUE)
evsr$just[evsr$type=="RWmean"]<-weighted.mean(evsNR$group,evsNR$propweight, na.rm=TRUE)
evsr$just[evsr$type=="NRmean"]<-mean(evsNR$groupM, na.rm=TRUE)

evsr$stick[evsr$type=="RUmean"]<-mean(evsNR$f5, na.rm=TRUE)
evsr$stick[evsr$type=="RWmean"]<-weighted.mean(evsNR$f5,evsNR$propweight, na.rm=TRUE)
evsr$stick[evsr$type=="NRmean"]<-mean(evsNR$v186, na.rm=TRUE)

write.table(evsr,file= "evsr.txt",sep="\t") 

#significance
#create the right dataset
evsbothw<-evsNR[which(evsNR$mainsurv==1),]
evsbothw$propweight2<-evsbothw$propweight
evsbothw$type<-1
evsbothw$type2<-1

evsbothnw<-evsbothw
evsbothnw$propweight<-1
evsbothnw$propweight2<-0
evsbothnw$type<-0
evsbothnw$type2<-NA

evsNRnw<-evsNR
evsNRnw$propweight<-0
evsNRnw$propweight2<-1
evsNRnw$type<-NA
evsNRnw$type2<-0

evssign<-rbind(evsbothw,evsbothnw,evsNRnw)
table(evssign$type2)
library(survey)
design1<-svydesign(ids=~1,weights=evssign$propweight, data=evssign)
svymean(evssign$impwork_h,design=design1, na.rm=TRUE)
svychisq(~type+impwork_h, design = design1, statistic = "F") 
svychisq(~type+impfami_h, design = design1, statistic = "F") 
svychisq(~type+impfrie_h, design = design1, statistic = "F") 
svychisq(~type+impleas_h, design = design1, statistic = "F")
svychisq(~type+imppol_h, design = design1, statistic = "F")
svychisq(~type+imprel_h, design = design1, statistic = "F")
svychisq(~type+trust_h, design = design1, statistic = "F")
svychisq(~type+immig_h, design = design1, statistic = "F")
svychisq(~type+just_h, design = design1, statistic = "F")
svychisq(~type+stick_h, design = design1, statistic = "F")
svychisq(~type+polintr_h, design = design1, statistic = "F")
#all p-value above .95
design2<-svydesign(ids=~1,weights=evssign$propweight2, data=evssign)
svymean(evssign$impwork_h,design=design2, na.rm=TRUE)
svychisq(~type2+impwork_h, design = design2, statistic = "F") 
svychisq(~type2+impfami_h, design = design2, statistic = "F") 
svychisq(~type2+impfrie_h, design = design2, statistic = "F") 
svychisq(~type2+impleas_h, design = design2, statistic = "F")
svychisq(~type2+imppol_h, design = design2, statistic = "F")
svychisq(~type2+imprel_h, design = design2, statistic = "F")
svychisq(~type2+trust_h, design = design2, statistic = "F")
svychisq(~type2+immig_h, design = design2, statistic = "F")
svychisq(~type2+just_h, design = design2, statistic = "F")
svychisq(~type2+stick_h, design = design2, statistic = "F")
svychisq(~type2+polintr_h, design = design2, statistic = "F")