R_code_Pu.R

## Meta_final_code
## by Pu
## 2024May2


library(metafor)
library(readxl)
metafin<-read_xlsx('C:/Users/cp/Desktop/Meta Analysis/Dataset_Pu.xlsx')

## z adjustment
metafin<-escalc("ZCOR",ri=ri, ni=ni,data=metafin,var.names = c('yzi','vzi'))

## zero effect size
randem_zero<- rma(yzi,vzi, data= metafin, slab = author) 
randem_zero

predict(randem_zero, digits=3, transf = transf.ztor)
confint(randem_zero)

## funnel plot
funnel(randem_zero)

## forest plot
forest(randem_zero, xlim = c(-1.2,1.2),atransf=transf.ztor,
       digits = c(2,1),
       cex=.6)
text(-1.2,24,"Authors(year)",pos=4, cex=1)
text(1.2,24,'r[95% CI]',pos=2,cex=1)

## small study effect test
regtest(randem_zero)

## influential case test
inf<- influence(randem_zero)
print(inf)
plot(inf)

## contour funnel plot
metafor::funnel(randem_zero, level=c(90, 95, 99), shade=c("white", "gray55", "gray75"), refline=0,
                xlab= "r",legend=TRUE)

## number of authors
summary(metafin$nauthor)
sd(metafin$nauthor)

metafin$nau4<-0
metafin$nau4[metafin$nauthor>=4]<-1
table(metafin$nau4)

randem_nau4<- rma(yzi,vzi, mods= ~nau4,data= metafin, slab = author) 
randem_nau4

## moving constant
metafin$invernau4<-1-metafin$nau4
randem_invernau4<- rma(yzi,vzi, mods= ~invernau4,data= metafin, slab = author) 
randem_invernau4

## cite
summary(metafin$cite)
sd(metafin$cite)

randem_cite<-rma(yzi,vzi, mods= ~cite,data= metafin, slab = author) 
randem_cite

## survey length
summary(metafin$days)
sd(metafin$days,na.rm=TRUE)
randem_days<-rma(yzi,vzi, mods= ~days,data= metafin, slab = author) 
randem_days

metafin$tw<-NA
metafin$tw[metafin$days>=14]<-1
metafin$tw[metafin$days<14& metafin$days>0]<-0
summary(metafin$tw)

## interpolation
metafin$twitpl<-1
metafin$twitpl[metafin$days<14& metafin$days>0]<-0

randem_tw<-rma(yzi,vzi, mods= ~tw,data= metafin, slab = author) 
randem_tw

randem_twitpl<-rma(yzi,vzi, mods= ~twitpl,data= metafin, slab = author) 
randem_twitpl

## moving constant
metafin$invertwitpl<-1-metafin$twitpl
randem_invertwitpl<-rma(yzi,vzi, mods= ~invertwitpl,data= metafin, slab = author) 
randem_invertwitpl

## nationwide
metafin$nw<-0
metafin$nw[metafin$scale=="nationwide"]<-1

summary(metafin$nw)
sd(metafin$nw)
randem_nw<-rma(yzi,vzi, mods= ~nw,data= metafin, slab = author) 
randem_nw

## quota
metafin$quota<-0
metafin$quota[metafin$sampling=="online quota"]<-1
summary(metafin$quota)
sd(metafin$quota)
randem_quota<-rma(yzi,vzi, mods= ~quota,data= metafin, slab = author) 
randem_quota

## average age
metafin$age40<-0
metafin$age40[metafin$aveage<40]<-1
summary(metafin$aveage)
sd(metafin$aveage)
randem_age<- rma(yzi,vzi, mods= ~aveage,data= metafin, slab = author) 
randem_age

randem_age40<- rma(yzi,vzi, mods= ~age40,data= metafin, slab = author) 
randem_age40

## female percent
summary(metafin$femper)
sd(metafin$femper)
randem_femper<- rma(yzi,vzi, mods= ~femper,data= metafin, slab = author) 
randem_femper

## average education
summary(metafin$aveedu)
sd(metafin$aveedu,na.rm=TRUE)
randem_edu<- rma(yzi,vzi, mods= ~aveedu,data= metafin, slab = author) 
randem_edu

## EU/US nation
table(metafin$nation)

metafin$euus<-0
metafin$euus[metafin$nation=='Finland'|metafin$nation=='France'|metafin$nation=='UK'|
               metafin$nation=='Czech'|metafin$nation=='Poland'|metafin$nation=='US'|metafin$nation=='Italy'
             |metafin$nation=='Serbia'|metafin$nation=='Slovakia']<-1
randem_euus<-rma(yzi,vzi, mods= ~euus,data= metafin, slab = author) 
randem_euus
## moving constant
metafin$inveuus<-1-metafin$euus
randem_euus<-rma(yzi,vzi, mods= ~inveuus,data= metafin, slab = author) 
randem_euus

## number of controls
summary(metafin$ncontrol)
sd(metafin$ncontrol)
randem_ncon<-rma(yzi,vzi, mods= ~ncontrol,data= metafin, slab = author) 
randem_ncon

## make plots
predncon<-predict(randem_ncon, newmods = seq(0,30,1))
ncons<- 1/sqrt(metafin$vzi)
nconsize<- (ncons-min(ncons)/(max(ncons)-min(ncons)))/30
plot(metafin$ncontrol,metafin$yzi,xlab = 'Numbers of Controls',
     ylab = "Correlation Coefficient (R)",xlim = c(0,30),ylim=c(0,0.8),cex=nconsize)
lines(seq(0,30,1), predncon$pred, col = "navy")
lines(seq(0,30,1), predncon$ci.lb, lty = "dashed", col=" blue")
lines(seq(0,30,1), predncon$ci.ub, lty = "dashed", col=" blue")
abline(h=0,col='red')

## months since 2020.1
metafin$monnumber<-metafin$month+(metafin$year-2020)*12
summary(metafin$monnumber)
sd(metafin$monnumber,na.rm = TRUE)

## interpolation
metafin$monnumberitpl<-metafin$monnumber
metafin$monnumberitpl[which(is.na(metafin$monnumberitpl))]<-8.579
summary(metafin$monnumberitpl)

randem_mon<-rma(yzi,vzi, mods= ~monnumber,data= metafin, slab = author) 
randem_mon

randem_monitpl<-rma(yzi,vzi, mods= ~monnumberitpl,data= metafin, slab = author) 
randem_monitpl

## plot
prednmonitpl<-predict(randem_monitpl, newmods = seq(0,20,1))
ncons<- 1/sqrt(metafin$vzi)
nconsize<- (ncons-min(ncons)/(max(ncons)-min(ncons)))/30
plot(metafin$monnumber,metafin$yzi,xlab = 'Numbers of Month Since 2020 Jan.',
     ylab = "Correlation Coefficient (R)",xlim = c(0,20),ylim=c(0,0.8),cex=nconsize)
lines(seq(0,20,1), prednmonitpl$pred, col = "navy")
lines(seq(0,20,1), prednmonitpl$ci.lb, lty = "dashed", col=" blue")
lines(seq(0,20,1), prednmonitpl$ci.ub, lty = "dashed", col=" blue")
abline(h=0,col='red')

## type of conspiracy
table(metafin$xcontent)
metafin$xcontb<-factor(metafin$xcontent,
                       levels = c('Political interests','vaccine','mixed','Bioweapon/man made','Hidden truth','hoax'))
randem_xcontb<-rma(yzi,vzi, mods= ~xcontb,data= metafin, slab = author) 
randem_xcontb

predxcontp<-predict(randem_xcontb, newmods = c(0,0,0,0,0))
predxcontp$pred
predxcontp$ci.lb
predxcontp$ci.ub

predxcontv<-predict(randem_xcontb, newmods = c(1,0,0,0,0))
predxcontv$pred
predxcontv$ci.lb
predxcontv$ci.ub

predxcontm<-predict(randem_xcontb, newmods = c(0,1,0,0,0))
predxcontm$pred
predxcontm$ci.lb
predxcontm$ci.ub

predxcontb<-predict(randem_xcontb, newmods = c(0,0,1,0,0))
predxcontb$pred
predxcontb$ci.lb
predxcontb$ci.ub

predxcontht<-predict(randem_xcontb, newmods = c(0,0,0,1,0))
predxcontht$pred
predxcontht$ci.lb
predxcontht$ci.ub

predxcontho<-predict(randem_xcontb, newmods = c(0,0,0,0,1))
predxcontho$pred
predxcontho$ci.lb
predxcontho$ci.ub

table(metafin$xcontent)

metafin$xcontnum<-0
metafin$xcontnum[metafin$xcontent=='Political interests']<-1
metafin$xcontnum[metafin$xcontent=='vaccine']<-2
metafin$xcontnum[metafin$xcontent=='mixed']<-3
metafin$xcontnum[metafin$xcontent=='Bioweapon/man made']<-4
metafin$xcontnum[metafin$xcontent=='Hidden truth']<-5
metafin$xcontnum[metafin$xcontent=='hoax']<-6

## plot
plot(metafin$xcontnum,metafin$yzi,xlab = 'Type of Conspiracy',
     ylab = "Correlation Coefficient (R)",ylim=c(-0.3,1.1),cex=nconsize)
legend('topleft',cex=0.8,
       c('1=Political interests','2=Vaccine','3=Mixed','4=Bioweapon/man made','5=Hidden truth','6=Hoax'))
lines(c(1,1,1),c(predxcontp$ci.lb,predxcontp$pred,predxcontp$ci.ub),lty=2,col='grey')
lines(c(2,2,2),c(predxcontv$ci.lb,predxcontv$pred,predxcontv$ci.ub),lty=2,col='grey')
lines(c(3,3,3),c(predxcontm$ci.lb,predxcontm$pred,predxcontm$ci.ub),lty=2,col='grey')
lines(c(4,4,4),c(predxcontb$ci.lb,predxcontb$pred,predxcontb$ci.ub),lty=2,col='grey')
lines(c(5,5,5),c(predxcontht$ci.lb,predxcontht$pred,predxcontht$ci.ub),lty=2,col='grey')
lines(c(6,6,6),c(predxcontho$ci.lb,predxcontho$pred,predxcontho$ci.ub),lty=2,col='grey')
points(1,predxcontp$pred,pch=1,lwd=5,col='grey')
points(2,predxcontv$pred,pch=1,lwd=5,col='grey')
points(3,predxcontm$pred,pch=1,lwd=5,col='grey')
points(4,predxcontb$pred,pch=1,lwd=5,col='grey')
points(5,predxcontht$pred,pch=1,lwd=5,col='grey')
points(6,predxcontho$pred,pch=1,lwd=5,col='grey')
abline(h=0,col='red')