Another plot for a reddit question

religion reddit.R

###Simple regression ####

#Read.csv just turns a csv into a data-frame and I assign it to the name "religion" w/
# the <- command.  Your file path will be different, of course.  I cleaned up the 
#data in google refine first, removing a wierd leading character in the 
#country column and the % sign in the religion column
religion<-read.csv(file="~/Downloads/Religion.csv",as.is=TRUE)

#assigns names to the columns
names(religion)<- c("country", "hdi", "religion", "murder")

#Simple linear regression of murder rates on religion, no controls.
rel.lm<- lm(murder ~ religion, data=religion)

#Controlling for HDI
cont.lm <- lm(murder ~ religion + hdi, data=religion)

#To get the results of either of these regressions, type
# summary(cont.lm) at the console.  

##Plots

#these two lines plot the relationship between religiosity and murder, note that even 
#though there appears to be a very slightly positive slope there is NO way to reject the
#possibility of no relationship whatsoever.
with(religion, plot(murder ~ religion))
abline(a=coef(rel.lm)[[1]], b=coef(rel.lm)[[2]], col=3, lty=2)

#we can also plot hdi vs. religion
with(religion, plot(hdi, religion))
#I wouldn't model this with a linear relationship (partially because both the dependent and explanatory 
#variables are bounded between 0 and 1, but also because the relationship looks non-linear
#but you can run a simple regression w/ this command
lm(religion ~ hdi, data=religion)