library(lattice)

data <- read.table(file="C:/Documents and Settings/steeleb/Desktop/DonnerParty.txt",head=T)
head(data)
n <- dim(data)[1]

dotplot(data$Age~data$Outcome|data$Sex,col="black",pch=16,ylab="Age")

table(data$Sex,data$Outcome)

Old <- data$Age > 29
table(data$Sex,data$Outcome,Old)
mosaicplot(table(data$Sex,data$Outcome,Old))
bwplot(data$Outcome~data$Age)
bwplot(data$Sex~data$Age)

####################################### formal comparison age differences between sexes #####################


TF.vector <- data$Sex=="FEMALE"

n.true <- sum(TF.vector)   
n.false <- n - n.true
mean.age.females <- mean(data$Age[TF.vector==T])
mean.age.males <- mean(data$Age[TF.vector==F])


c(n.true,mean.age.females)
c(n.false,mean.age.males)


######################## simulate sample means assuming no difference between females and males ##########

n.sims <- 10000
simulated.means <- rep(0,n.sims)

for (i in 1:n.sims){
	random.sample <- sample(1:n,size=n.true)
	simulated.means[i] <- mean(data$Age[random.sample])
}
plot(density(simulated.means),main="")
x <- sum(simulated.means <= mean.age.females)
c(n.sims,x)
mean(simulated.means<=mean.age.females)

######## Poisson approx
lambda <- n.sims/2
ppois(x,lambda=lambda)

############################# Homework ###############
# 1. Compare ages between survivors and those that died, and state whether there is 
# clear evidence that those that died tended to be older than the survivors. Support your
# position with a graph and the results from the simulation
# 2. The mortality rate among men is substantially greater than that among women. This result may be
# attributed to gender differences or perhaps human behavior. What is your explanation?