#### Quant 1
#### Week 4 Code

require(foreign)
dail2002 <- read.dta("../02 Univariate/dail2002.dta")
dail2002$incumbf <- factor(dail2002$incumb,
                           labels=c("Challenger", "Incumbent"))
dail2002$wonseatf <- factor(dail2002$wonseat,
                            labels=c("Lost", "Won"))

## incumbency by outcome
table(dail2002$incumbf, dail2002$wonseatf)

## incumbency by outcome by gender
levels(dail2002$gender) <- c("Male", "Female")
table(dail2002$incumbf, dail2002$wonseatf, dail2002$gender)
ftable(table(dail2002$incumbf, dail2002$wonseatf, dail2002$gender), col.vars=c(2,3))
ftable(table(dail2002$incumbf, dail2002$wonseatf, dail2002$gender), col.vars=c(3,2))

## two-way tables
partyinctable <- table(dail2002$party, dail2002$incumbf)
addmargins(partyinctable)     # both row and column marginals
addmargins(partyinctable, 2)  # row marginals
addmargins(partyinctable, 1)  # column marginals
options("digits"=1)
prop.table(partyinctable)      # proportion of all cells
prop.table(partyinctable, 1)   # row proportions
prop.table(partyinctable, 2)   # column proportions
addmargins(prop.table(partyinctable))        # total proportion marginals
addmargins(prop.table(partyinctable, 1), 2)  # row proportions and sum
addmargins(prop.table(partyinctable, 2), 1)  # column proportions and sum

## three-way tables
(partyincgendertable <- table(dail2002$party, dail2002$incumbf, dail2002$gender))
ftable(partyincgendertable)

## boxplots to compare a continuous variable by categories of a factor variable
attach(dail2002)
temp.medians <- aggregate(spend_total, by=list(party), median, na.rm=TRUE)
temp.partyf  <- factor(party, levels=temp.medians[order(temp.medians$x), 1])
boxplot(spend_total ~ temp.partyf, ylab="Total Spending in Euros")
levels(temp.partyf) <- toupper(levels(temp.partyf))
boxplot(spend_total ~ temp.partyf, varwidth=T, ylab="Total Spending in Euros")
detach(dail2002)


## illustration of data frame indexing, and problems with using attach()
is.na(dail2002$votes1st)
sum(is.na(dail2002$votes1st))
dail2002$wholename[is.na(dail2002$votes1st)]
dail2002[is.na(dail2002$votes1st), c("wholename","votes1st")]

attach(dail2002)
votes1st[is.na(votes1st)] <- -99    # change this missing value code
cbind(wholename, votes1st)[379,]    # look at local copy
cbind(dail2002$wholename, dail2002$votes1st)[379,] # compare to dframe
detach(dail2002)