# R-program for Supplementary Exercise 1.145 of IPS7e

csdata <- read.csv("h:/vhm/vhm801/Datasets/csv/Chapter 1/ex01_145.csv")

# descriptive statistics
summary(csdata[csdata$sex==1,])
summary(csdata[csdata$sex==2,])
# can also use doBy library to compute statistics by groups (here, sex)
library(doBy)
summaryBy (gpa+satm ~ sex, csdata, FUN=c(mean,sd,median,IQR,length))

# boxplots
boxplot (gpa ~ sex, csdata)
boxplot (satm ~ sex, csdata)
# Stata note for gpa boxplot outliers applies to R implementation as well

# quantile plots
# note: datax=T option give plot orientation as in Minitab, otherwise as in Stata
qqnorm(csdata[csdata$sex==1,]$gpa, datax=T)
qqline(csdata[csdata$sex==1,]$gpa, datax=T)
# same thing for sex=2 (women)
qqnorm(csdata[csdata$sex==2,]$gpa, datax=T)
qqline(csdata[csdata$sex==2,]$gpa, datax=T)
# and now for satm
qqnorm(csdata[csdata$sex==1,]$satm, datax=T)
qqline(csdata[csdata$sex==1,]$satm, datax=T)
qqnorm(csdata[csdata$sex==2,]$satm, datax=T)
qqline(csdata[csdata$sex==2,]$satm, datax=T)

# Shapiro-Wilk normality test
shapiro.test(csdata[csdata$sex==1,]$gpa)
shapiro.test(csdata[csdata$sex==2,]$gpa)
shapiro.test(csdata[csdata$sex==1,]$satm)
shapiro.test(csdata[csdata$sex==2,]$satm)
# note: not exactly same P-values as in Stata