# Assignment 2

# Options

# Surpress scientific notations, digits and lines

options(scipen=999, digits=2, max.print=99999999)

# MEAN Scores

attach(DELL_REL)

DELL_REL$FAM<-(FAM01 + FAM02 + FAM03)/3
DELL_REL$SN<-(SN01 + SN02 + SN03 + SN04)/4
DELL_REL$SA<-(SA01 + SA02 + SA03 + SA04)/4
DELL_REL$TR<-(TR01 + TR02 + TR03 + TR04)/4
DELL_REL$RC<-(RC01 + RC02 + RC03 + RC04)/4
DELL_REL$RQ<-(RQ01 + RQ02 + RQ03)/3
DELL_REL$LOY<-(LOY01 + LOY02 + LOY03)/3
DELL_REL$WOM<-(WOM01 + WOM02 + WOM03)/3


# Correlation

cor(DELL_REL[,2:29])

cor(DELL_REL[,35:42])

library(Hmisc)

rcorr(as.matrix(DELL_REL[,35:42]))

R<-rcorr(as.matrix(DELL_REL[,35:42]))
str(R)
print(R$r)
print(R$P)


library(corrgram)

corrgram(DELL_REL[,2:29])

corrgram(DELL_REL[,35:42])


# Sample Size

library(pwr)

# R2=0.30
# u=3 (independent variables)
# n=v+u+1

pwr.f2.test(u=3, f2=0.05/(1-0.05), sig.level=0.05, power=0.80)

pwr.f2.test(u=6, f2=0.25/(1-0.250), sig.level=0.05, power=0.80)

pwr.f2.test(u=6, f2=0.25/(1-0.250), sig.level=0.005, power=0.80)


# Simple Regression

R.01<-lm(RC ~ TR)
summary(R.01)

plot(RC,TR)
abline(lm(RC ~ TR))


# Cook's D plot

plot(R.01, which=4)


# Hierarchical Model

R.02<-lm(RC ~ TR + FAM + SN + SA)
summary(R.02)

anova(R.01,R.02)

library(car)

vif(R.02)


# Variable Importance

library(QuantPsyc)
lm.beta(R.02)

library(relaimpo)
calc.relimp(R.02, rela=TRUE)


# Demographics

attr(GENDER, "labels")
attr(EDUCATION, "labels")

R.03<-lm(RC ~ TR + FAM + SN + SA + factor(GENDER) + AGE + factor(EDUCATION))
summary(R.03)

anova(R.01,R.02,R.03)


# Testing Linear Hypothesis

R.10<-lm(RQ ~ RC + TR)
summary(R.10)

library(car)

linearHypothesis(R.10, "RC=TR")


# Core Model

R.21<-lm(LOY ~ RQ)
summary(R.21)

R.22<-lm(WOM ~ RQ)
summary(R.22)


# Mediation

library(psych)

MED.1<-mediate(LOY ~ TR + RC + (RQ), data=DELL_REL, n.iter=5000)
mediate.diagram(MED.1, digits=10)
print(MED.1, digits=10, short=FALSE)

MED.2<-mediate(WOM ~ TR + RC + (RQ), data=DELL_REL, n.iter=5000)
mediate.diagram(MED.2, digits=10)
print(MED.2, digits=10, short=FALSE)

# Moderation 

library(psych)

MOD1<-mediate(LOY ~ RQ*LENGTH, data=DELL_REL)
print(MOD1, digits=6)

MOD2<-mediate(LOY ~ RQ*INCOME, data=DELL_REL)
print(MOD2, digits=6)

MOD3<-mediate(WOM ~ RQ*LENGTH, data=DELL_REL)
print(MOD3, digits=6)

MOD4<-mediate(WOM ~ RQ*INCOME, data=DELL_REL)
print(MOD4, digits=6)


# jtools

library(jtools)

MOD1.1<-lm(LOY ~ RQ*LENGTH, data=DELL_REL)
summary(MOD1.1)

johnson_neyman(model = MOD1.1, pred = RQ, modx = LENGTH)