House & presidential election returns

The R data package uspoliticalextras includes federal election returns from a couple of sources, namely the Daily Kos and the MIT Election Data and Science Lab. I use the package as a bit of a cache for US political data; it is available via Git Hub. The focus here will be on the last Presidential election (2016) & the last two House races (2016 & 2018).

library(tidyverse)
#devtools::install_github("jaytimm/uspoliticalextras")  
pres <- uspoliticalextras::uspol_dkos_returns_pres_cd 
house <- uspoliticalextras::uspol_medsl_returns_house_cd %>%
  filter(year > 2015) %>%
  mutate(party = ifelse(party == 'Independent', 'Republican Party', party)) # Amash

Trump margins in 2016

Number of congressional districts carried by DJT & HRC.

pres %>%
  filter(year == 2016) %>%
  group_by(candidate) %>%
  count() %>%
  knitr::kable()

Trump margins by congressional district in 2016.

library(sf)
to_map <- uspoliticalextras::uspol_dkos_equalarea_sf$hex_cds %>%
  select(GEOID) %>%
  left_join(pres %>% 
              filter(year == 2016) %>% 
              mutate(trump_margin = republican - democrat),
            by = 'GEOID') 
to_map %>%
  ggplot() + 
  geom_sf(aes(fill = trump_margin),
          color = 'white') +
  geom_sf(data=uspoliticalextras::uspol_dkos_equalarea_sf$hex_states, 
          fill = NA, 
          show.legend = F, 
          color="#5a5c5b", 
          lwd=.5) +
  ggsflabel::geom_sf_text(data = uspoliticalextras::uspol_dkos_equalarea_sf$hex_cds,
                          aes(label = district_code), 
                          size = 2,
                          color='black') +
  scale_fill_distiller(palette = "RdBu", direction=-1) +
  theme_minimal()+
  theme(axis.text.x=element_blank(),
        axis.text.y=element_blank(),
        axis.title.x=element_blank(),
        axis.title.y=element_blank(),
        legend.position = 'none') +
  labs(title = "2016 Trump Margins by Congressional District",
       caption = 'Source: Daily Kos')

Flipped House Districts from 2016 to 2018

Congressional districts that elected a Republican representative in 2016 and a Democratic representative in 2018.

house_flips <- house %>%
  select(GEOID, congress,party) %>%
  spread(congress, party) %>%
  left_join(house %>% 
              filter(year == 2018) %>% 
              mutate(house_rep_margin = round(republican - democrat, 1)) %>%
              select(GEOID, state, district_code, candidate, house_rep_margin) %>% 
              rename(house_rep = candidate))

Some corrections/amendments for Pennsylvania districts per 2019 redistricting.

pa16_rs <- c('4201', '4205', '4206',
           '4207', '4209', '4210',
           '4211', '4212', '4213','4215',
           '4216', '4217')
pa16_ds <- c('4202', '4203', '4204',
           '4208',  '4214',
           '4218')
pa18_rs <- c('4203', '4204', '4205', '4218')
pa18_ds <- c('4213', '4214')
house_flips1 <-  house_flips %>%
  mutate(`115` = ifelse(GEOID %in% pa16_rs, 'Republican Party', `115`),
         `115` = ifelse(GEOID %in% pa16_ds, 'Democratic Party', `115`)) %>%
  mutate(house_flip = paste0(`115`, ' | ', `116`)) %>%
  left_join(to_map %>% 
              sf::st_drop_geometry() %>% 
              select(GEOID, candidate, trump_margin)) %>%
  rename(Pres16 = candidate) %>%
  mutate(`116` = ifelse(GEOID %in% pa18_rs, 'Republican Party', `116`),
         `116` = ifelse(GEOID %in% pa18_ds, 'Democratic Party', `116`)) %>%  
  mutate(Pres16_House18 = paste0(Pres16, ' | ', `116`))

Democrats netted a total of 40 seats in the midterm elections in 2018. The numbers & districts presented below align with those presented on Bollotpedia.

house_flips1 %>%
  group_by(house_flip) %>%
  count() %>%
  knitr::kable()

House Representatives from flipped districts:

house_flips1 %>%
  filter(house_flip %in% 
           c('Democratic Party | Republican Party', 
             'Republican Party | Democratic Party')) %>%
  mutate(house_flip = ifelse(grepl('^D', house_flip), 'D -> R', 'R -> D')) %>%
  select(house_flip, state, district_code, house_rep) %>%
  DT::datatable(rownames = FALSE)

The 31 House Democrats in Trump-supportive districts

The table below summarizes how districts voted in the 2016 presidential election and House elections in 2018. Again, 31 House Democrats represent congressional districts that Trump won in 2016’s presidential election. In contrast, only three Republicans represent districts carried by HRC. Note: Numbers & districts align with those presented in this article from The Hill.

house_flips1 %>%
  group_by(Pres16_House18) %>%
  count() %>%
  #mutate(n = formattable::color_tile('white', 'steelblue')(n)) %>%
  knitr::kable()

The 31 Democratic lawmakers representing Trump won districts include:

house_flips1 %>%
  filter(Pres16_House18 == 'Trump | Democratic Party') %>%
  select(state, district_code, house_rep) %>%
  DT::datatable(rownames = FALSE)

A quick geographical perspective

to_map2 <- uspoliticalextras::uspol_dkos_equalarea_sf$hex_cds %>%
  select(GEOID) %>%
  left_join(house_flips1, by = 'GEOID') 

to_map2 %>%
  ggplot() + 
  geom_sf(aes(fill = Pres16_House18),
          alpha = .75,
          color = 'white') +
  geom_sf(data=uspoliticalextras::uspol_dkos_equalarea_sf$hex_states, 
          fill = NA, 
          show.legend = F, 
          color="black", 
          lwd=.5) +
  ggsflabel::geom_sf_text(data = uspoliticalextras::uspol_dkos_equalarea_sf$hex_cds,
                          aes(label = district_code), 
                          size = 2,
                          color='black') +
  ggthemes::scale_fill_stata() +
  theme_minimal()+
  theme(axis.text.x=element_blank(),
        axis.text.y=element_blank(),
        axis.title.x=element_blank(),
        axis.title.y=element_blank(),
        legend.position = 'bottom') +
  labs(title = "2016 Presidential Support & 2018 House Representative",
       subtitle = 'By Congressional District')

The 13 House Democrats in solid Trump districts

Of the 31, 13 Democrats represent districts Trump carried by more than 6 points.

house_flips1 %>%
  filter(Pres16_House18 == 'Trump | Democratic Party' & trump_margin > 6) %>%
  select(state, district_code, house_rep, trump_margin, house_rep_margin) %>%
  arrange(desc(trump_margin)) %>%
  knitr::kable()

Voting patterns in presidential elections

Counts of how districts voted in the last three presidential elections are presented below.

house_flips2 <- house_flips1 %>%
  left_join (pres %>% 
               group_by(GEOID) %>%
               summarize(pres_lineage = paste0(candidate, collapse = ' | ')) %>%
               ungroup())
house_flips2 %>%
  group_by(pres_lineage) %>%
  count() %>%
  knitr::kable()

Voting patterns for the 31 Trump-House Dem districts

12 out of the 31 Dem-Trump districts have voted for Republican presidential candidates in the last three elections, ie, McCain-Romney-Trump districts.

house_flips2 %>%
  filter(Pres16_House18 == 'Trump | Democratic Party') %>%
  group_by(pres_lineage) %>%
  count() %>%
  knitr::kable()

Representatives for these twelve districts are presented below.

house_flips2 %>%
  filter(Pres16_House18 == 'Trump | Democratic Party' & 
           pres_lineage == 'McCain | Romney | Trump') %>%
  select(state, district_code, house_rep) %>%
  knitr::kable()

The 5 House Democrats that should probably vote against impeachment

The table below lists the five House Reps representing districts that have supported Republican presidential candidates in the last three elections, with 2016 Trump margins greater than ten points.

house_flips2 %>%
  filter(Pres16_House18 == 'Trump | Democratic Party' & 
           trump_margin > 10 &
           pres_lineage == 'McCain | Romney | Trump') %>%
  select(state, district_code, house_rep, pres_lineage, trump_margin) %>%
  arrange(desc(trump_margin)) %>%
  knitr::kable()

Front-line Freshmen House Democrats

Via the Rvoteview package, we identify 88 freshman House Representatives, 56 of which are Democrats.

house_vv_freshies <- Rvoteview:: member_search(chamber= 'House', congress = 116) %>%
  filter(congresses == 'c(116, 116)' & party_name != 'Independent') %>%
  select(bioname, party_name) %>%
  rename(house_rep = bioname) 

house_vv_freshies %>%
  group_by(party_name) %>%
  count() %>%
  knitr::kable()

Of the 31, then, 22 are freshman. So, a pretty vulnerable group.

house_flips3 <- house_flips2 %>%
  left_join(house_vv_freshies %>% 
              mutate(is_fresh = 'Freshman') %>% 
              select(house_rep, is_fresh)) %>%
  mutate(is_fresh = ifelse(is.na(is_fresh), 'Not', is_fresh))

house_flips3 %>%
  group_by(is_fresh, Pres16_House18) %>%
  count() %>%
  #filter(is_fresh == 'Y') %>%
  spread(is_fresh, n) %>%
  janitor::adorn_totals(where = c('row', 'col')) %>%
  knitr::kable()

Data set

The data set compiled in this post/guide is available here.