# Options

# Surpress scientific notations, digits and lines

options(scipen=999, digits=10, max.print=99999999)


# One Way Anova (Completely Randomized Design)

attach(Table16.2)
# detach() # To detach data set ...

FIT.01<- aov(sales ~ factor(promotion))

# FIT.01<-aov(sales ~ factor(promotion), data=Table16.2)


library(lsr)
etaSquared(FIT.01, type=3, anova=TRUE)

summary(FIT.01)

print(summary(FIT.01), digits=10)


print(model.tables(FIT.01,"means"),digits=10) 

# Type III === SPSS
# To get SPSS equivalent SPSS results

options(contrasts=c("contr.sum", "contr.poly"))

drop1(FIT.01,~.,test="F")

# Install.packages("car") is R Base

library(car)

Anova(FIT.01, type=3)

pairwise.t.test(sales, factor(promotion), p.adj = "bonf")

TukeyHSD(FIT.01)




# Regression Approach

FIT.LM.01<-lm(sales ~ factor(promotion))
summary(FIT.LM.01)
anova(FIT.LM.01)

library(car)
Anova(FIT.LM.01, type=3)





# Descriptives and Diagnostics

print(model.tables(FIT.01,"means"),digits=3)

library(psych)
describeBy(sales, factor(promotion))

library(lawstat)
levene.test(sales,factor(promotion), location="mean")

library(car)
leveneTest(sales,factor(promotion), center="mean")

# bartlett.test(sales ~ factor(promotion))




# Plots

plot(FIT.01)

library(gplots)

boxplot(sales ~ factor(promotion))

plotmeans(sales ~ factor(promotion), xlab="Promotion",
          ylab="$", main="Mean Plot with 95% CI") 



# Welch Correction

oneway.test(sales ~ factor(promotion))




# Kruskal-Wallis Test

kruskal.test(sales ~ factor(promotion))

pairwise.wilcox.test(sales,factor(promotion), p.adj="bonf")

library(coin)

kruskal_test(sales ~ factor(promotion), distribution="approximate")




# Two-way ANOVA

FIT.02<- aov(sales ~ factor(promotion)*factor(coupon))

summary(FIT.02)

print(model.tables(FIT.02,c("means"),digits=3))

library(lsr)
etaSquared(FIT.02, type=3, anova=TRUE)


# Type III === SPSS
# To get SPSS equivalent SPSS results

options(contrasts=c("contr.sum", "contr.poly"))

drop1(FIT.02,~.,test="F")

#library(car)
Anova(FIT.02, type=3)


# Regression Model

FIT.LM.02<-lm(sales ~ factor(promotion)*factor(coupon))
summary(FIT.LM.02)
anova(FIT.LM.02)

#library(car)
Anova(FIT.LM.02, type=3)

plot(FIT.02)

interaction.plot(factor(promotion),factor(coupon),sales) 

# ANCOVA

FIT.03<- aov(sales ~ clientel + factor(promotion)*factor(coupon))

summary(FIT.03)
library(car)
Anova(FIT.03, type=3)



# Easypower

library(easypower)


# Defining Effect Size
# Etasquared s 0.01, m 0.06, l 0.138

main.eff.1 <- list(name = "Promotion", levels = 3, eta.sq = 0.06)
main.eff.2 <- list(name = "Coupon", levels = 2, eta.sq = 0.06)

# Running n.multiway

n.multiway(iv1=main.eff.1, iv2=main.eff.2, interaction.eta2 = 0.01, sig.level = 0.05, power = 0.8)