CY Leung ratings, 201403

# Date file was retrieved on March 14, 2014, from
# http://hkupop.hku.hk/chinese/popexpress/ce2012/dataset/tp1403013.sav

############################
# To load in the data file #
############################

# Install the "foreign" package, to read in SPSS files
install.packages("foreign")

# Load in the "foreign" package
library("foreign")

# Read in the data file
cy201403 <- read.spss("tp1403013.sav", to.data.frame=T)

# Check if file has been read in correctly
head(cy201403)


######################
# Basic descriptions #
######################

# Number of subjects = 1017
dim(cy201403)[1]

# Number of raters = 998
sum(table(cy201403$CE_rating))



##################
# On raw ratings #
##################

# A quick scan of the distribution of the raw ratings
table(cy201403$CE_rating)

# Mean of raw ratings = 47.4008
mean(cy201403$CE_rating, na.rm=T)

# Median of raw ratings = 50
median(cy201403$CE_rating, na.rm=T)

# Histogram of the raw ratings
hist(cy201403$CE_rating, main="CY Leung's rating (raw), \nMarch 2013",
    xlab="Raw Scores", ylab="Frequency", right=F, axes=F,
    breaks=seq(0, 100, 10), col="lightgray")
axis(1, at=seq(0, 100, 10))
axis(2, at=seq(0, 300, 50))

# Indicate mean on plot
abline(v=mean(cy201403$CE_rating, na.rm=T), col="red", lwd=4)

# Indicate median on plot
abline(v=median(cy201403$CE_rating, na.rm=T), col="blue", lwd=4)

# Add labels
text(mean(cy201403$CE_rating, na.rm=T), 200, "Mean", col="red", lwd=4, pos=2)
text(median(cy201403$CE_rating, na.rm=T), 200, "Median", col="blue", lwd=4, pos=4)


######################
# On weighed ratings #
######################

# Calculate weighed ratings
cy201403$weighedRating <- cy201403$CE_rating * cy201403$weight

# Mean of raw ratings = 47.48411
mean(cy201403$weighedRating, na.rm=T)

# Median of raw ratings = 41.9557
median(cy201403$weighedRating, na.rm=T)

# Histrogram of the weighed ratings
hist(cy201403$weighedRating, main="CY Leung's rating (weighed), \nMarch 2013",
    xlab="Weighed Scores", ylab="Frequency", right=F, axes=F,
    breaks=seq(0, 500, 10), col="lightgray")
axis(1, at=seq(0, 500, 50))
axis(2, at=seq(0, 140, 20))

# Indicate mean on plot
abline(v=mean(cy201403$weighedRating, na.rm=T), col="red", lwd=4)

# Indicate median on plot
abline(v=median(cy201403$weighedRating, na.rm=T), col="blue", lwd=4)

# Add labels
text(mean(cy201403$weighedRating, na.rm=T), 100, "Mean", col="red", lwd=4, pos=4)
text(median(cy201403$weighedRating, na.rm=T), 100, "Median", col="blue", lwd=4, pos=2)