DATA101-basicpoliticaldataanalysis

#Question 1A

# Getting county_returns.csv

library(tidyverse)

install.packages("maps")
install.packages("mapdata")

library(maps)
library(mapdata)
library(readr)
county_returns <- read_csv("~/Documents/Education/Dropbox/DATA101/DATA/RAW/county_returns.csv")

# Merging county-level mapping data with electoral data

county_votes  <- counties%>%
  full_join(county_returns, by = c("region" = "state.name"))
  full_join(county_returns, by = c("subregion" = "Geographic.Name"))
  
#1B - Creating new variables and integrating them with election/county data.
  
clinton_trump_outcomes <- subset(county_votes, select = -c(16, 14, 13, 12))%>%
  mutate(clinton_prop = clinton/total_votes_2016)%>%
  mutate(trump_prop = trump/total_votes_2016)%>% 
  mutate(clinton_won = ifelse(trump_clinton_margin < 0, 'Yes', 'No'))

#Question 2: Mapping the data

library(ggplot2)
library(viridisLite) #(No viridis package found)
library(scales)

# Clinton v Trump (County and state-level)

ggplot() +
  geom_polygon(data=clinton_trump_outcomes, aes(x = long, y = lat, group = group, fill = clinton_prop, color = clinton_won))+
  coord_quickmap()+
  theme_void()+
  ggtitle("2016 Results")

# Question 3

#County-level analysis:

ggplot() +
  geom_polygon(data=clinton_trump_outcomes, aes(x = long, y = lat, group = group, fill = clinton_prop))+
  scale_fill_gradient2(low = muted("red"), mid = "white", high = muted("blue"), midpoint = .5)+
  coord_quickmap()+
  theme_void()+
  ggtitle("2016 Election Results")+
  labs(fill = "% Dem Vote")

#State-level analysis:

ggplot() +
  geom_polygon(data=states, aes(x=long, y=lat, group=group), fill = NA, col="lightgray", lwd=0.25)+
  geom_polygon(data=clinton_trump_outcomes, aes(x = long, y = lat, group = group, fill = clinton_won))+
  coord_quickmap()+
  theme_void()+
  ggtitle("2016 Election Results")+
  labs(fill = "Clinton Won")

#3A

ggplot() +
  geom_polygon(data=clinton_trump_outcomes, aes(x = long, y = lat, group = group, fill = clinton_prop))+
  scale_fill_gradient(low = "red", high = "blue")+
  coord_quickmap()+
  theme_void()+
  ggtitle("2016 Results")+
  labs(fill = "% Clinton Won")

#Though the state level option does well to immediately help you pinpoint 
# who took the majority of a state, I think it important to leverage the gradient
# when possible, as it offers a much more contextual perspective on how a greater number of
# individuals participated in our election. Seeing this map paints a much clearer picture
# of how Trump won back then, because seeing a statewide view of where Clinton may have won 
# the majority still didn't accurately describe the sentiments of the non-liberal counties.

# Question 4

# To analyze Obama's margin over Clinton:

# The data:

romney_obama_outcomes <- subset(county_votes, select = -c(9, 10, 11, 15))%>%
  mutate(obama_prop = obama/total_votes_2012)%>%
  mutate(romney_prop = romney/total_votes_2012)%>% 
  mutate(obama_won = ifelse(romney_obama_margin < 0, 'Yes', 'No'))

# Obama v. Romney:

ggplot() +
  geom_polygon(data=romney_obama_outcomes, aes(x = long, y = lat, group = group, fill = obama_prop))+
  scale_fill_gradient(low = "red", high = "blue")+
  coord_quickmap()+
  theme_void()+
  ggtitle("2012 Results")+
  labs(fill = "% Obama Won")

# Obama v. Clinton:


clinton_obama_outcomes <- subset(county_votes, select = -c(9, 13,15, 16))%>%
  mutate(obama_prop2 = obama/total_votes_2012)%>%
  mutate(clinton_prop2 = clinton/total_votes_2016)%>% 
  mutate(clinton_success = ifelse(clinton_prop2 > obama_prop2, 'Better', 'Worse'))

# On a map:

ggplot() +
  geom_polygon(data=clinton_obama_outcomes, aes(x = long, y = lat, group = group, fill = clinton_prop2))+
  scale_fill_gradient(low = "red", high = "blue")+
  coord_quickmap()+
  theme_void()+
  ggtitle("2012 v 2016")+
  labs(fill = "% Clinton > Obama")

# Comparatively speaking, Clinton success was primarily in any state where a major city accounted
# for the majority of the state's population (though one could argue
# things were more polarized in 2016, which cut her campaign off at the knees). Obama (somehow) 
# not only captured the coasts,but had a lot more support the midwest and rockies;
# places where Clinton had a hard time capturing a single county. In all fairness, 
# however, the candidates of the 2012 election were much more 
# moderate overall, and that notion is reflected by the fact that over 2/3 of the 
# states were a shade of purple in 2012.