pres_results data off

[brshallo]

Results look different than expected. For example https://en.wikipedia.org/wiki/2016_United_States_presidential_election_in_Washington_(state) says 2016 election had 54.3% of vote for Clinton and 38.07% for Trump. Or perhaps I'm misunderstanding specifics of data. Also the total for dem, rep, and other does not add-up to 1.

library(tidyverse)

politicaldata::pres_results %>% 
  filter(year == 2016, state == "WA")
#>   year state total_votes       dem       rep      other
#> 1 2016    WA     3317019 0.5253868 0.3683268 0.07378583

^{Created on 2020-09-10 by the reprex package (v0.3.0)}

[brshallo]

Looked into a little closer and checked data source, and difference is caused by whether write-in votes are counted or not.

[brshallo]

To get a similar output to yours but with vote proportions that come to zero I installed .csv data from data source and then ran:

library(tidyverse)

vote_raw <- read_csv("1976-2016-president.csv")

votes_clean <- vote_raw %>% 
  mutate(party = ifelse(is.na(party), "", party)) %>% 
  mutate(party = case_when(party == "democrat" ~ "dem",
                           party == "republican" ~ "rep",
                           TRUE ~ "other")) %>% 
  # filter(!writein) %>% # include if not wanting to include write-ins
  select(-state) %>% 
  rename(state = state_po) %>% 
  group_by(year, state, party) %>% 
  summarise(votes = sum(candidatevotes)) %>% 
  ungroup() %>% 
  spread(party, votes) %>% 
  relocate(other, .after = last_col()) %>% 
  mutate(across(c(dem, rep, other), ~ ifelse(is.na(.), 0, .))) %>% 
  mutate(total_votes = dem + rep + other) %>% 
  mutate(across(c(dem, rep, other), ~ . / total_votes))

See brshallo/weighted-national-popular-vote.