Voter Shift

registered-voters.R

name,registered_voters_2020,registered_voters_2016
Aitkin,"10,840","10,090"
Anoka,"228,495","206,600"
Becker,"21,396","18,829"
Beltrami,"27,518","24,213"
Benton,"24,688","22,106"
Big Stone,"3,212","3,090"
Blue Earth,"39,978","36,184"
Brown,"16,209","15,052"
Carlton,"22,253","20,424"
Carver,"70,539","61,884"
Cass,"20,559","18,421"
Chippewa,"7,321","6,858"
Chisago,"36,747","32,855"
Clay,"36,387","32,465"
Clearwater,"5,030","4,605"
Cook,"4,039","3,653"
Cottonwood,"6,531","6,297"
Crow Wing,"44,215","39,427"
Dakota,"283,710","257,792"
Dodge,"12,812","11,776"
Douglas,"26,440","24,302"
Faribault,"8,475","8,262"
Fillmore,"13,042","12,221"
Freeborn,"18,629","17,920"
Goodhue,"30,732","28,320"
Grant,"3,985","3,852"
Hennepin,"835,366","759,075"
Houston,"12,415","12,026"
Hubbard,"14,147","12,620"
Isanti,"25,822","22,926"
Itasca,"29,104","27,110"
Jackson,"6,266","5,985"
Kanabec,"10,060","9,264"
Kandiyohi,"25,592","23,719"
Kittson,"2,854","2,730"
Koochiching,"7,614","7,169"
Lac Qui Parle,"4,392","4,226"
Lake,"7,595","7,206"
Lake of the Woods,"2,641","2,443"
Le Sueur,"18,040","16,499"
Lincoln,"3,465","3,392"
Lyon,"14,322","13,559"
McLeod,"22,605","20,667"
Mahnomen,"2,883","2,483"
Marshall,"5,602","5,306"
Martin,"12,115","11,805"
Meeker,"14,321","13,493"
Mille Lacs,"15,884","14,442"
Morrison,"21,104","19,085"
Mower,"21,677","20,042"
Murray,"5,381","5,085"
Nicollet,"21,025","19,988"
Nobles,"10,066","9,536"
Norman,"3,824","3,615"
Olmsted,"101,371","90,259"
Otter Tail,"39,096","35,742"
Pennington,"7,777","7,400"
Pine,"16,959","15,526"
Pipestone,"5,482","5,370"
Polk,"17,628","16,516"
Pope,"7,375","6,897"
Ramsey,"334,758","305,584"
Red Lake,"2,331","2,157"
Redwood,"8,974","9,355"
Renville,"8,784","8,508"
Rice,"39,508","35,912"
Rock,"5,549","5,244"
Roseau,"9,260","8,543"
St. Louis,"131,653","123,666"
Scott,"94,317","82,598"
Sherburne,"60,396","53,667"
Sibley,"9,290","8,522"
Stearns,"93,373","86,457"
Steele,"22,912","21,026"
Stevens,"5,537","5,753"
Swift,"5,638","5,567"
Todd,"13,982","13,007"
Traverse,"2,003","2,002"
Wabasha,"14,248","13,277"
Wadena,"8,198","7,539"
Waseca,"11,222","10,580"
Washington,"178,499","159,431"
Watonwan,"5,682","5,866"
Wilkin,"3,799","3,583"
Winona,"29,674","28,825"
Wright,"87,354","75,963"
Yellow Medicine,"5,970","5,924"

voter_shift.R

library(tidyverse)
library(jsonlite)
library(rvest)
library(plotly)
library(glue)

# 2020 Data ---------------------------------------------------------------

nyt <- "https://static01.nyt.com/elections-assets/2020/data/api/2020-11-03/state-page"
df <- fromJSON(glue("{nyt}/minnesota.json"))$data$races$counties[[1]]

# Voter Registration 2016/2020  -------------------------------------------

registered <- read_csv("registered-voters.csv")
webpage <- read_html("https://www.nytimes.com/elections/2016/results/minnesota")
tbls <- html_nodes(webpage, "table")
# Shows all tables where Walz is matched
df_2016 <- tbls %>% 
  grep("Hennepin",.) %>% # Find table with a county
  tbls[.] %>% 
  html_table(fill=T) %>% .[[1]] %>% # Convert table to DF
  rename(name = `Vote by county`) %>%
  filter(!name %>% is.na) %>% # Drop NA
  arrange(name) %>% # Get to line up with NYT 2020 data
  mutate(Clinton = Clinton %>% str_replace('[^0-9]','') %>% as.numeric,
         Trump = Trump %>% str_replace('[^0-9]','') %>% as.numeric)

# Contains the county names, 2016 registered voters, and 2020 registered voters

results <- df %>% left_join(registered) %>% arrange(name) # Ordered by County


# Build Votes DF ----------------------------------------------------------

# DF for plot
votes <- data.frame(
  county = results$name,
  dem_2020 = results$results$bidenj,
  rep_2020 = results$results$trumpd,
  dem_2016 = df_2016$Clinton,
  rep_2016 = df_2016$Trump,
  voters_2020 = results$registered_voters_2020,
  voters_2016 = results$registered_voters_2016,
  rep_2016_percent = df_2016$Trump / results$registered_voters_2016 * 100,
  dem_2016_percent = df_2016$Clinton / results$registered_voters_2016 * 100,
  rep_2020_percent = results$results$trumpd / results$registered_voters_2020 * 100,
  dem_2020_percent = results$results$bidenj / results$registered_voters_2020 * 100
)
votes$dem_2016_offset <- votes$dem_2016_percent - mean(votes$rep_2016_percent %>% append(votes$dem_2016_percent))
votes$rep_2016_offset <- votes$rep_2016_percent - mean(votes$rep_2016_percent %>% append(votes$dem_2016_percent))

votes$dem_2020_offset <- votes$dem_2020_percent - mean(votes$rep_2016_percent %>% append(votes$dem_2016_percent))
votes$rep_2020_offset <- votes$rep_2020_percent - mean(votes$rep_2016_percent %>% append(votes$dem_2016_percent))


# Plot Data ---------------------------------------------------------------

p1 <- ggplot(votes) + 
  geom_segment(aes(x = dem_2016_offset, y = rep_2016_offset, 
                   xend = dem_2020_offset, yend = rep_2020_offset, 
                   colour = county,
                   size = voters_2020),
               arrow = arrow(length = unit(0.03, "npc"))) +
  geom_abline(slope=1, intercept=0) +
  scale_x_continuous(limits = c(-25,30)) +
  scale_y_continuous(limits = c(-25,30)) +
  theme(legend.position = "none") +
  xlab("Democratic") +
  ylab("Republican")
ggsave(p1,filename = "plot.png",width=10,height=10)