options(max.print=1000000)


#load packages and database----------------

library(haven)
library(dplyr)
library(ggrepel)
library (ggplot2)
library(plyr)
library (ggthemes)
library (sjmisc)
library (stargazer)
library(car)
library (psych)
library (GPArotation)
library (lavaan)
library (semPlot)
library (lm.beta)
library (semTools)
library (polycor)
library (summarytools)
library (officer)
library (flextable)
library (tidyr)


df1 <- read_spss ("korona_data_clean.sav")

#database without participants with missing data
df1nm <- df1[which(df1$missing == 0),]


#CFA's-------------------

#COVID-19 unfounded beliefs=====================

consp_2f1hfinal <- '
PODR =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
LIJEC=~ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001

UB=~ a*PODR+ a*LIJEC 
'
fit.consp_2f1hfinal<- cfa (consp_2f1hfinal, data=df1, estimator ="ML"
                         #                         , missing ="ML"
)
summary (fit.consp_2f1hfinal, fit.measures =T, standardized = T)
reliability(fit.consp_2f1hfinal)

consp_1ffinal <- '
CONSp =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
+ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001
'
fit.consp_1ffinal<- cfa (consp_1ffinal, data=df1, estimator ="ML"
                         #                         , missing ="ML"
)
summary (fit.consp_1ffinal, fit.measures =T, standardized = T)

anova (fit.consp_1ffinal, fit.consp_2f1hfinal)

# CRT==============
crt_1f<-'
CRT =~ crttn1 + crttn2 + crttn3
'

fit.crt_1f<- cfa (crt_1f, df1, estimator="ML"
#                   , missing ="ML"
)
summary (fit.crt_1f, fit.measures=T, standardized=T)





#AOT i Faith in intuition======================

aot_2f <- '
AOT =~ aot1SQ001 + aot2SQ001 + aot3SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + aot9SQ001 + aot10SQ001 
FI =~ fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001
'
fit.aot_2f<- cfa (aot_2f, data=df1, estimator ="ML"
#                  , missing ="ML"
                  )
summary (fit.aot_2f, fit.measures =T, standardized = T)

#without items 3 and 9

aot_2f <- '
AOT =~ aot1SQ001 + aot2SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + 
aot10SQ001 
FI =~ fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001
'
#AOTh=~ 1*AOT+ -1*FI'
fit.aot_2f<- cfa (aot_2f, data=df1, estimator ="ML"
                  #                  , missing ="ML"
)
summary (fit.aot_2f, fit.measures =T, standardized = T)

aot_1f <- '
AOT =~ aot1SQ001 + aot2SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + 
aot10SQ001 + fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001

'
fit.aot_1f<- cfa (aot_1f, data=df1, estimator ="ML"
                  #                  , missing ="ML"
)
summary (fit.aot_1f, fit.measures =T, standardized = T)

anova (aot_2f, aot_1f)


#COVID-19 behavior================

pon_1f<-'
PONASANJE =~ ponasanje1SQ001 + ponasanje2SQ001 + ponasanje3SQ001 + ponasanje4SQ001'

fit.pon_1f <- cfa(pon_1f, data = df1, estimator = "ML" ) 
summary(fit.pon_1f, fit.measures = TRUE, standardized = TRUE)



#science curiosity=============

scicur_1f<-'
SCICUR =~ scicur1SQ006 + scicur3SQ006 + scicur4SQ006 + scicur5SQ006 + scicur6SQ006
'

fit.scicur_1f <- cfa(scicur_1f, data = df1, estimator = "ML" ) 
summary(fit.scicur_1f, fit.measures = TRUE, standardized = TRUE)

#anxiety=================

anx_1f<- '
ANX =~ anx1SQ001 + anx2SQ001 + anx3SQ001 + anx4SQ001
'

fit.anx_1f <- cfa(anx_1f, data = df1, estimator = "ML" ) 
summary(fit.anx_1f, fit.measures = TRUE, standardized = TRUE)



#measurement model with all scales============================

#all scales on separate factor

sve<-'
#predictors
CRT =~ crttn1 + crttn2 + crttn3
AOT =~ aot1SQ001 + aot2SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + aot10SQ001 
FI =~ fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001
SCICUR =~ scicur1SQ006 + scicur3SQ006 + scicur4SQ006 + scicur5SQ006 + scicur6SQ006
ANX =~ anx1SQ001 + anx2SQ001 + anx3SQ001 + anx4SQ001

#mediators
PODR =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
LIJEC=~ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001

UB=~ a*PODR+ a*LIJEC 

#criterion
PONASANJE =~ ponasanje1SQ001 + ponasanje2SQ001 + ponasanje3SQ001 + ponasanje4SQ001
'

fit.sve <- cfa(sve, data = df1, estimator = "ML" ) 
summary(fit.sve, fit.measures = TRUE, standardized = TRUE)

#predictors on two cognitive and anxiety factors

sve_p2f<-'
#predictors
KOGN =~ crttn1 + crttn2 + crttn3+ 
aot1SQ001 + aot2SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + aot10SQ001 
+fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001
+scicur1SQ006 + scicur3SQ006 + scicur4SQ006 + scicur5SQ006 + scicur6SQ006
ANX =~ anx1SQ001 + anx2SQ001 + anx3SQ001 + anx4SQ001

#mediators
PODR =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
LIJEC=~ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001

UB=~ a*PODR+ a*LIJEC 

#criterion
PONASANJE =~ ponasanje1SQ001 + ponasanje2SQ001 + ponasanje3SQ001 + ponasanje4SQ001
'


fit.sve_p2f <- cfa(sve_p2f, data = df1, estimator = "ML" ) 
summary(fit.sve_p2f, fit.measures = TRUE, standardized = TRUE)

anova (fit.sve, fit.sve_p2f)


#descriptive statistics of mean scale scores-------------------------
#anx=anxiety scale, crt= cognitive reflection test, aot = actively open-minded thinking scale without item 3 and 9,
#povj_intuic=faith in intuition, sci_cur= science curiosity, zabrSQ001 = COVID-19 worry, consp= COVID-19 Conspiracy Beliefs,
#lijecenje= COVID-19 Treatment Beliefs, UB= COVID-19 Averaged Unfounded Beliefs, znanje= COVID-19 Knowledge, 
#ponasanje = COVID-19 Behavior 
#spol=sex, dob = age, obrazovanje = education,ideologijaSQ001 = political values

desk<- describe(df1[, c("anx", "crt", "aotclean", "povj_intuic", "sci_cur", "zabrSQ001", "consp", "lijecenje",
                        "UB", "znanje", "ponasanje", "spol", "dob", "obrazovanje", "ideologijaSQ001")])

dfSummary(df1[, c("anx", "crt", "aotclean", "povj_intuic", "sci_cur", "zabrSQ001", "consp", "lijecenje",
                  "UB", "znanje", "ponasanje", "spol", "dob", "obrazovanje", "ideologijaSQ001")])



## correlations among latent factors==============


#rescaling age to equalize it's variance with other variables' variance

df1$dobr<- (df1$dob)/10

korel<-'
#predictors
CRT =~ crttn1 + crttn2 + crttn3
AOT =~ aot1SQ001 + aot2SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + aot10SQ001 
FI =~ fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001
SCICUR =~ scicur1SQ006 + scicur3SQ006 + scicur4SQ006 + scicur5SQ006 + scicur6SQ006
ANX =~ anx1SQ001 + anx2SQ001 + anx3SQ001 + anx4SQ001


#mediators
PODR =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
LIJEC=~ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001

UB=~ a*PODR+ a*LIJEC 

ZNANJE=~ znanje
ZABR=~ zabrSQ001

#criterion
PONASANJE =~ ponasanje1SQ001 + ponasanje2SQ001 + ponasanje3SQ001 + ponasanje4SQ001

#control varibles (obrazovanje = education,ideologijaSQ001 = political values) 

SEX=~ spol
AGE=~ dobr
EDU=~ obrazovanje
IDE=~ ideologijaSQ001


'

fit.korel <- cfa (korel, data = df1, estimator = "ML" ) 
summary(fit.korel, fit.measures = TRUE, standardized = TRUE)
parameterestimates(fit.korel, standardized = T)
reliability(fit.korel)

#test of the scalar invariance and effect of consider the opposite intervention (uputa=consider the opposite intervention,
#1=intervetion group, 2=control group) =================


consp_2f1hfinal_int <- '
PODR =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
LIJEC=~ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001

UB=~ 1*PODR+ 1*LIJEC 
'
fit.consp_2f1hfinal_int<- cfa (consp_2f1hfinal_int, data=df1, estimator ="ML", group= "uputa",
                               meanstructure=T
                               #                         , missing ="ML"
)
summary (fit.consp_2f1hfinal_int, fit.measures =T, standardized = T)

fitci.consp_2f1hfinal_int<- cfa (consp_2f1hfinal_int, data=df1, estimator ="ML", group= "uputa",
                                 meanstructure=T, group.equal = c("intercepts", "loadings")
                                 #                         , missing ="ML"
)
summary (fitci.consp_2f1hfinal_int, fit.measures =T, standardized = T)

fitcm.consp_2f1hfinal_int<- cfa (consp_2f1hfinal_int, data=df1, estimator ="ML", group= "uputa",
                                 meanstructure=T, group.equal = c("means", "intercepts", "loadings")
                                 #                         , missing ="ML"
)
summary (fitcm.consp_2f1hfinal_int, fit.measures =T, standardized = T)

anova (fit.consp_2f1hfinal_int, fitci.consp_2f1hfinal_int, fitcm.consp_2f1hfinal_int)


#Behavior prediction SEM model-------------------


#model without control variables

modelnc <- '
#preditcors
CRT =~ crttn1 + crttn2 + crttn3
AOT =~ aot1SQ001 + aot2SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + aot10SQ001 
FI =~ fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001
SCICUR =~ scicur1SQ006 + scicur3SQ006 + scicur4SQ006 + scicur5SQ006 + scicur6SQ006
ANX =~ anx1SQ001 + anx2SQ001 + anx3SQ001 + anx4SQ001


#mediators
PODR =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
LIJEC=~ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001

UB=~ a*PODR+ a*LIJEC 

#criterion
PONASANJE =~ ponasanje1SQ001 + ponasanje2SQ001 + ponasanje3SQ001 + ponasanje4SQ001

#regressions
UB~ acu*CRT+ asu*SCICUR+ aau*AOT+ afu*FI+ aanu*ANX
znanje~  aczn*CRT+ aszn*SCICUR+ aazn*AOT+ afzn*FI+ aanzn*ANX
zabrSQ001~ acza*CRT+ asza*SCICUR+ aaza*AOT+ afza*FI+ aanza*ANX
PONASANJE~ bu*UB+ bzn*znanje+ bzab*zabrSQ001+ cc*CRT+ cs*SCICUR+ ca*AOT+ cf*FI+ can*ANX


#covariances

UB~~ znanje+ zabrSQ001
znanje~~ zabrSQ001

#indirect effects
crtub := acu*bu
aotub := aau*bu
fiub := afu*bu

anxzab := aanza*bzab
'
fit.modelnc <- sem(modelnc, data = df1, estimator = "ML" 
                                  , se = "boot", bootstrap = 5000 
                  #                  ,missing="ML"
                   #                 , parallel = "snow", ncpus= 3, verbose= T
) 

summary(fit.modelnc, fit.measures = TRUE, standardized = TRUE)

parameterestimates(fit.modelnc, standardized = T, boot.ci.type = "bca.simple"
                   , level = 0.95) %>% .[136:142,]

inspect(fit.modelnc, 'rsquare')

#model with control variables

#rescaling age to equalize it's variance with other variables' variance

df1$dobr<- (df1$dob)/10


model <- '
#predictors
CRT =~ crttn1 + crttn2 + crttn3
AOT =~ aot1SQ001 + aot2SQ001 + aot4SQ001 + aot5r + aot6SQ001 + aot7r + aot8r + aot10SQ001 
FI =~ fi1SQ001 + fi2SQ001 + fi3SQ001 + fi4SQ001 + fi5SQ001
SCICUR =~ scicur1SQ006 + scicur3SQ006 + scicur4SQ006 + scicur5SQ006 + scicur6SQ006
ANX =~ anx1SQ001 + anx2SQ001 + anx3SQ001 + anx4SQ001


#mediators
PODR =~ podrijetlo1SQ001 + podrijetlo2SQ001 + podrijetlo3SQ001 + podrijetlo4SQ001 + podrijetlo5SQ001
+ ozbiljnost1SQ001+ ozbiljnost2SQ001+ ozbiljnost3SQ001
LIJEC=~ lijecenje1SQ001 + lijecenje2SQ001 + lijecenje3SQ001 + lijecenje4SQ001 + lijecenje5SQ001 + 
lijecenje6SQ001

UB=~ a*PODR+ a*LIJEC 

#criterion
PONASANJE =~ ponasanje1SQ001 + ponasanje2SQ001 + ponasanje3SQ001 + ponasanje4SQ001

#regressions
UB~ acu*CRT+ asu*SCICUR+ aau*AOT+ afu*FI+ aanu*ANX
znanje~  aczn*CRT+ aszn*SCICUR+ aazn*AOT+ afzn*FI+ aanzn*ANX
zabrSQ001~ acza*CRT+ asza*SCICUR+ aaza*AOT+ afza*FI+ aanza*ANX
PONASANJE~ bu*UB+ bzn*znanje+ bzab*zabrSQ001+ cc*CRT+ cs*SCICUR+ ca*AOT+ cf*FI+ can*ANX


#control varibles (spol = sex, obrazovanje = education,
#ideologijaSQ001 = political values

UB~ spol+ dobr+ obrazovanje+ ideologijaSQ001
znanje~ spol+ dobr+ obrazovanje+ ideologijaSQ001
zabrSQ001~ spol+ dobr+ obrazovanje+ ideologijaSQ001
PONASANJE~ spol+ dobr+ obrazovanje+ ideologijaSQ001

#covariances

UB~~ znanje+ zabrSQ001
znanje~~ zabrSQ001

#indirect effects
crtub := acu*bu
aotub := aau*bu
fiub := afu*bu

anxzab := aanza*bzab
'
fit.model <- sem(model, data = df1, estimator = "ML"
                 , se = "boot", bootstrap = 5000 
                 #                  ,missing="ML"
                           , parallel = "snow", ncpus= 3, verbose= T
) 

summary(fit.model, fit.measures = TRUE, standardized = TRUE)

parameterestimates(fit.model, standardized = T, boot.ci.type = "bca.simple"
                   , level = 0.99, rsquare = T) %>% .[c(46:84,162:168),]

inspect(fit.model, 'rsquare')