Does ballot position matter for City Council?

This May, Philadelphia will be voting for City Council. This includes five city-wide Democratic At Large positions. We don’t yet know exactly how many At Large candidates there will be, but in 2019 there were 28 names on the ballot.

In order to arrange those names on the ballot, we famously draw names from a coffee can.

In the past, I’ve demonstrated that our judicial elections are determined by the random luck of drawing a good ballot position: being in the first column nearly triples your votes, and is more important than a Democratic City Committee endorsement and Philadelphia Bar Association Recommendation combined). I even proposed an NBA-wheel style ballot procedure that would fix the problem.

I’ve wondered if the same effect exists for City Council. There are reasons to expect not: voters pay more attention to City Council races and candidates spend more money, so it’s less likely that a voter will just push a button in the first column. But with voters choosing up to five names out of a pool of around 28 candidates, it’s certainly plausible they’ll take shortcuts.

I tried this analysis in January 2019 but didn’t have quite enough data. This time around I’ve added in 2019’s 28 candidates, and can finally measure some effects.

In 2019, all three incumbents plus Isaiah Thomas won handily. The fifth winner was Katherine Gilmore Richardson with 6.8% of the vote. Following her were Justin DiBerardinis with 6.3%, then Adrián Rivera-Reyes, Eryn Santamoor, and Erika Almirón at 5.3, 5.2, and 5.1% respectively.

View code
library(tidyverse)
library(sf)

source("../../admin_scripts/util.R")


setwd("C:/Users/Jonathan Tannen/Dropbox/sixty_six/posts/council_ballot_position_23/")
df_major <- readRDS("../../data/processed_data/df_major_type_20220523.Rds")
ballot_position <- read.csv("../../data/processed_data/ballot_layout.csv")

Encoding(ballot_position$candidate) <- "latin1"
ballot_position$candidate <- gsub("\\s+", " ", ballot_position$candidate)

format_name <- function(x){
  x <- tolower(x)
  x <- gsub("(\\b)([a-z])", "\\1\\U\\2", x, perl=TRUE)
  x <- gsub("(á|ñ|ó)([A-Z]+)", "\\1\\L\\2", x, perl=TRUE)
  x <- gsub("\\s+", " ", x)
  x <- gsub("(^\\s)|(\\s$)", "", x)
  return(x)
}

council <- df_major %>% 
  filter(
    election_type == "primary",
      party == "DEMOCRATIC",
      office == "COUNCIL AT LARGE",
      year %in% c(2011, 2015, 2019)
  ) %>%
  mutate(year = as.integer(year))

council <- council %>% 
  left_join(ballot_position, by = c("year" = "year", "candidate" = "candidate"))

council$candidate <- factor(council$candidate)
levels(council$candidate) <- format_name(levels(council$candidate))
council <- council %>% filter(candidate != 'Write In')

council <- council %>%
  group_by(year) %>%
  mutate(ncand = length(unique(candidate)))

total_results <- council %>%
  group_by(candidate, year, row, column, ncand, incumbent) %>%
  summarise(votes = sum(votes)) %>%
  group_by(year) %>%
  mutate(
    pvote = votes/sum(votes),
    winner = rank(desc(votes)) <= 5
  )

YEAR <- 2019
ggplot(
  total_results %>% 
    filter(year == YEAR) %>% 
    mutate(
      lastname=format_name(gsub(".*\\s(\\S+)$", "\\1", candidate)),
      lastname=ifelse(lastname == "Jr",format_name(gsub(".*\\s(\\S+\\s\\S+)$", "\\1", candidate)),lastname),
    ) %>%
    arrange(votes),
  aes(y=row, x=column)
) +
  geom_tile(
    aes(fill=pvote*100, color=winner),
    size=2
  ) +
  geom_text(
    aes(
      label = ifelse(incumbent==1, "Incumbent", ""),
      x=column-0.45,
      y=row+0.45
    ),
    color="grey70",
    hjust=0, vjust=0
  ) +
  geom_text(
    aes(label = sprintf("%s\n%0.1f%%", lastname, 100*pvote)),
    color="black"
    # fontface="bold"
  ) +
  scale_y_reverse(NULL) +
  scale_x_continuous(NULL)+
  scale_fill_viridis_c(guide=FALSE) +
  scale_color_manual(values=c(`FALSE`=rgb(0,0,0,0), `TRUE`="yellow"), guide=FALSE) +
  expand_limits(x=3.5)+
  theme_sixtysix() %+replace% 
  theme(
    panel.grid.major=element_blank(),
    axis.text=element_blank()
  ) +
  ggtitle(
    paste(YEAR, "Council At Large Results"),
    "Democratic Primary, arranged by the ballot layout. Winners are outlined."
  )

Ballot position appears weaker than for judges: many candidates win from later columns. Incumbency is obviously the strongest factor.

But looking farther back, we see instances where ballot position appears to help. In 2015, Derek Green led the entire field as a challenger with the top position. And in 2011 Sherrie Cohen came in a close sixth place from the first column, and two more first column candidates were in the top nine.

View code
ggplot(
  total_results,
  aes(y = 100 * pvote, color = interaction(incumbent, column==1))
) + 
  geom_text(
    aes(label = candidate),
    x=0, 
    hjust=0
  ) +
  facet_grid(. ~ year) +
  theme_sixtysix() +
  scale_y_continuous(breaks = seq(0,20,2.5)) +
  geom_text(
    data = tribble(
      ~votes, ~candidate, ~incumbent, ~year, ~pvote, ~column,
      # 1e3, "Challenger", 0, 2011, -0.007, 0,
      7e3, "Incumbent", 1, 2011, 0.007, 0,
      4e3, "First Column", 0, 2011, 0.000, 1
    ),
    fontface="bold",
    x=0.45,
    aes(label = candidate),
    hjust = 0,
    vjust=0
  ) +
  scale_color_manual(
    values=c(
      '1.FALSE' = strong_blue, 
      '1.TRUE' = strong_blue, 
      '0.FALSE'= "black", 
      '0.TRUE'=strong_green
    ),
    guide = FALSE
  ) +
  expand_limits(y=0) +
  labs(
    title="Incumbents Swept 2011 and 2019, but not 2015",
    y = "% of Vote"
  )

Let’s use regression to tease apart the effects. I’ll regress the percent of the vote received by a candidate on being in the first column and being in the first row, incumbency, and year fixed effects. The regression is simplistic, but since ballot position is randomized we don’t need anything more. (The substantive findings below are robust to more controls and to using log(votes).)

View code
ols_fit <- lm(
  100 * pvote ~ 
    as.character(year) +
    incumbent +
    (row == 1) +
    (column == 1) + 
    # (column == 1 & row == 1) +
    # (column == 1 & row != 1) +
    # (column == 2) +
    1,
  data = total_results #%>% filter(!incumbent)
)
# summary(ols_fit)

print_coef <- function(fit, coef){
  val <- round(ols_fit$coefficients[coef], 1)
  se <- summary(ols_fit)$coefficients[,2][coef]
  # stars <- case_when(p<0.01 ~ " (p < 0.01)", p < 0.05 ~ " (p < 0.05)", TRUE ~ "")
  se_text <- paste0(" (",round(se, 1),")")
  prefix <- (if(val > 0) "+" else "")
  paste0(prefix, val, se_text)
}

tribble(
  ~Effect, ~"% Vote in pp (standard error)",
  "Baseline Votes 2019", "2.6",
  "Incumbency", ols_fit %>% print_coef('incumbent'),
  "First Column",  ols_fit %>% print_coef('column == 1TRUE'),
  "First Row",  ols_fit %>% print_coef('row == 1TRUE')
) %>% 
  knitr::kable("html") %>% 
  kableExtra::kable_styling(full_width = F)
Effect % Vote in pp (standard error)
Baseline Votes 2019 2.6
Incumbency +6.4 (0.9)
First Column +2.4 (0.9)
First Row -0.3 (0.8)

Non-incumbent candidates in the second or later column started with an average 2.4% of the vote in 2019. Incumbents on average receive 6.4 percentage points more votes. Candidates in the first column receive on average 2.4 pp more votes. Being in the first row doesn’t appear to help.

So the first-column effect for City Council is smaller than for Common Pleas, but still nearly doubles a typical challenger’s votes. And in 2019 it would have been enough to put any of the close challengers (Gilmore Richardson, DiBerardinis, Santamoor, Almirón) over the top.

Common Pleas Deep Dive, 2021

Belatedly, I’ve had time to sit down with the 2021 Primary results. Here are some observations.

In November, Philadelphia will elect eight new judges on the Court of Common Pleas. After the May Primary, we know almost certainly who those judges will be; the Democratic nominees will all win.

All eight Democratic nominees are Recommended by the Bar, three Highly. Surprisingly, they don’t include the person in the number one ballot position. And they won with a wide diversity of maps.

View code
library(dplyr)
library(tidyr)
library(ggplot2)
devtools::load_all("../../admin_scripts/sixtysix/")

ballot <- read.csv("../../data/common_pleas/judicial_ballot_position.csv")
res <- readxl::read_xlsx("C:/Users/Jonathan Tannen/Downloads/2021_primary (1).xlsx")
res <- res %>%
  pivot_longer(
    cols=`JUSTICE OF THE\r\nSUPREME COURT DEM\r\nMARIA MCLAUGHLIN`:`QUESTION #5\r\nNO`,
    names_to="candidate",
    values_to="votes"
  )
names(res) <- gsub("(\\r|\\n)+", " ", names(res))
names(res) <- gsub("\\s", "_", tolower(names(res)))

res$vote_type <- case_when(
  res$vote_type == "E" ~ "Election Day",
  res$vote_type == "M" ~ "Mail",
  res$vote_type == "P" ~ "Provisional"
)

res_cp <- res %>%
  filter(
    grepl("^JUDGE OF THE\r\nCOURT OF COMMON PLEAS DEM\r\n", candidate)
  ) %>%
  mutate(
    candidate = gsub("^JUDGE OF THE\r\nCOURT OF COMMON PLEAS DEM\r\n","", candidate)
  )

res_cp <- res_cp %>%
  left_join(ballot %>% filter(year == 2021) %>% mutate(candidate = toupper(name))) %>%
  mutate(name=format_name(name))

assertthat::assert_that(
  res_cp %>% filter(is.na(name)) %>% with(all(candidate == "Write-in"))
)

res_type <- res_cp %>%
  filter(!is.na(name)) %>%
  group_by(name, vote_type, rownumber, colnumber, philacommrec, dcc, inq) %>%
  summarise(votes=sum(votes)) %>%
  group_by(vote_type) %>%
  mutate(pvote = votes/sum(votes))
View code
res_total <- res_type %>% 
  group_by(name, rownumber, colnumber, philacommrec, dcc, inq) %>%
  summarise(votes=sum(votes), .groups="drop") %>%
  mutate(pvote = votes/sum(votes))

ggplot(
  res_total %>% arrange(votes) %>% mutate(winner = rank(-votes) <= 8),
  aes(y=rownumber, x=colnumber)
) +
  geom_tile(
    aes(fill=pvote*100, color=winner),
    size=2
  ) +
  geom_text(
    aes(
      label = ifelse(philacommrec==1, "R", ifelse(philacommrec==2,"HR","")),
      x=colnumber+0.45,
      y=rownumber+0.45
    ),
    color="grey70",
    hjust=1, vjust=0
  ) +
  geom_text(
    aes(
      label = ifelse(dcc==1, "D", ""),
      x=colnumber-0.45,
      y=rownumber+0.45
    ),
    color="grey70",
    hjust=0, vjust=0
  ) +
  geom_text(
    aes(label = sprintf("%s\n%0.1f%%", name, 100*pvote)),
    color="black"
    # fontface="bold"
  ) +
  scale_y_reverse(NULL) +
  scale_x_continuous(NULL)+
  scale_fill_viridis_c(guide=FALSE) +
  scale_color_manual(values=c(`FALSE`=NA, `TRUE`="yellow"), guide=FALSE) +
  annotate(
    "text",
    label="R = Recommended\nHR = Highly Recommended\nD = DCC Endorsed",
    x = 1.6,
    y = 6,
    hjust=0,
    vjust=0.5,
    color="grey70"
  ) +
  theme_sixtysix() %+replace% 
  theme(
    panel.grid.major=element_blank(),
    axis.text=element_blank()
  ) +
  ggtitle(
    "Common Pleas Results",
    "2021 Democratic Primary, arranged as on the ballot. Winners are outlined."
  )

Four candidates won in the first column, three in the second, and one in the third. Three winners were Highly Recommended by the Bar, including Michele Hangley in the second column and Chris Hall in the third, but by itself that rating wasn’t sufficient: John Padova and Mark Moore failed to capitalize on it. There’s some additional work needed to use it to your advantage.

The candidates’ maps are diverse. Nick Kamau and Cateria McCabe won everywhere, though slightly stronger in the Black wards of West and North Philly (and decidedly not the Northeast). Wendi Barish also won everywhere, slightly stronger in Center City and its ring. Betsy Wahl, Chris Hall, and Michele Hangley all won thanks to their strength in the Wealthy Progressive ring around Center City and in Chestnut Hill and Mount Airy. Craig Levin did the opposite, winning the Northeast and West and North Philly, presumably on the strength of his DCC endorsement. And Dan Sulman was the eighth and final winner, with the bright yellow 53rd ward just enough to push him through, where his sister is the Ward Leader.

View code
library(sf)

divs <- st_read("../../data/gis/warddivs/202011/Political_Divisions.shp") %>%
  mutate(warddiv = pretty_div(DIVISION_N))wards <- st_read("../../data/gis/warddivs/201911/Political_Wards.shp") %>%
  mutate(ward=sprintf("%02d", asnum(WARD_NUM)))res_ward_type <- res_cp %>%
  mutate(ward = substr(division, 1, 2)) %>%
  group_by(ward, name, vote_type) %>%
  summarise(votes=sum(votes)) %>%
  group_by(vote_type) %>%
  mutate(pvote=votes/sum(votes))

res_ward <- res_ward_type %>%
  group_by(ward, name) %>%
  summarise(votes=sum(votes)) %>%
  group_by(ward) %>%
  mutate(pvote=votes/sum(votes))

res_ward <- left_join(wards, res_ward)

candidate_order <- res_total %>% arrange(desc(votes)) %>% with(name)

ggplot(
  res_ward %>% 
    filter(!is.na(name)) %>%
    mutate(name=factor(name, levels=candidate_order))
) +
  geom_sf(aes(fill=100*pvote), color=NA) +
  scale_fill_viridis_c("Vote %") +
  facet_wrap(~ name) +
  theme_map_sixtysix() %+replace% theme(legend.position="right") +
  ggtitle("Common Pleas Results", "2021 Democratic Primary")

Caroline Turner was the first runner up, and the first candidate to fail to win from the top ballot position since at least 2007 (which is all the ballot layouts I can find). But she did really well in the 1st and 2nd Wards, which now deserve a name.

The Reclaim Wards

When clicking through the results online, I saw a cut that made me laugh out loud.

Results from Division 01-01

The top eight winners in 01-01 each received more than 9.99% of the vote. Ninth place? Only 2.5%. This is the kind of electoral coordination party bosses dream of.

In fact, that consolidation is true of the entire first ward (covering East Passyunk in South Philly).

View code
ward_bar <- function(ward, endorsements){
  df <- res_ward %>%
  as.data.frame() %>%
  filter(ward==!!ward, !is.na(name)) %>%
  arrange(desc(votes)) %>%
  # filter(1:n() <= 10) %>%
  mutate(name=factor(name, levels=name)) %>%
  mutate(
    last_name = gsub(".* ([A-Za-z]+)$", "\\1", name),
    endorsed= last_name %in% endorsements,
    reclaim = last_name %in% c("Hall", "Hangley", "Kamau", "Barish","Sulman", "McCabe", "Turner", "Wahl")
  )
  
  ggplot(df, aes(x=name, y=votes)) +
  geom_bar(stat="identity", aes(color=endorsed, fill=reclaim), size=1.2) +
      geom_vline(xintercept=8.5, linetype="dashed") +
  scale_color_manual(
    NULL,
    values=c(`TRUE` = "goldenrod", `FALSE`=NA),
    labels=sprintf(c("Not %s Endorsed", "%s Endorsed"), ward)
  ) +
  scale_fill_manual(
    NULL,
    values=c(`TRUE`="grey30", `FALSE`="grey60"),
    labels=c("Not Reclaim", "Reclaim")
  ) +
  theme_sixtysix() %+replace% 
  theme(axis.text.x = element_text(angle=45, hjust=1, vjust=1))+
  labs(
    title=sprintf("Common Pleas Results in Ward %s", ward),
    subtitle="2021 Democratic Primary",
    x=NULL,
    y="Votes"
  ) 

}

ward_bar(
  "01", 
  c("Hall", "Hangley", "Kamau", "Barish","Sulman", "McCabe", "Turner", "Wahl")
)  

It’s usually impossible to separate the many overlapping endorsements. Was it the Bar that brought the win, the DCC, or the Ward? But these eight winners are exactly the ward’s endorsed candidates. They were also the full Reclaim slate, so it’s impossible to separate the Ward’s power from Reclaim. But the deciding factors were almost certainly these two.

The four wards with the biggest gap between eighth and ninth place–suggesting the strongest slate power–were South Philly’s 1st and 2nd and West Philly’s 27th and 46th.

The 2nd Ward, just to the 1st’s North, had slightly different endorsements than Reclaim. The six candidates who had both a Reclaim and 2nd Ward endorsement did best. Barish came in seventh with only a Reclaim endorsement, Levin in eighth with only the 2nd, and Sulman in ninth with only Reclaim.

View code
ward_bar(
  "02", 
  c("Hall", "Hangley", "Kamau", "Levin", "McCabe", "Turner", "Wahl")
)  

In the 27th (where, full disclosure, I’m a committeeperson), the Reclaim endorsements appear to have carried the day: Turner and Hall won, while 27-endorsed Moore and Levin didn’t.

View code
ward_bar(
  "27",
      c("Barish", "Moore", "Hangley", "Kamau", "Levin", "McCabe", "Sulman", "Wahl")
)

I haven’t found the 46th ward endorsements, but the Reclaim slate cleaned up.

View code
ward_bar(
  "46",
      c()
) + labs(subtitle="2021 Primary. Endorsements not available.")

Not only was the Reclaim slate particularly strong in these wards, but the gap between eighth and ninth position make it clear it was the Reclaim endorsement itself, and not one of the other pregressive slates that drove voters.

But the Reclaim endorsement wasn’t itself enough to win across the city. Caroline Turner came in ninth despite it and first ballot position, mostly due to poor results in the Black wards.

Wealthy Progressive divisions did consolidate their votes in a way other divisions didn’t. Grouping divisions by my Voting Blocs shows that the Wealthy Progressive divisions’ preferred candidates did better there than the preferred candidates of other blocs’.

View code
devtools::load_all("C:/Users/Jonathan Tannen/Dropbox/sixty_six/posts/svdcov/")
svd_time <- readRDS("../../data/processed_data/svd_time_20210813.RDS")
div_cats <- get_row_cats(svd_time, 2020)

# ggplot(divs %>% left_join(div_cats, by=c("warddiv"="row_id"))) +
#   geom_sf(aes(fill=cat), color=NA)

res_cat <- res_cp %>% left_join(div_cats, by=c("division"="row_id")) %>%
  filter(!is.na(name)) %>%
  group_by(name, cat) %>%
  summarise(votes=sum(votes)) %>%
  group_by(cat) %>%
  mutate(
    rank = rank(desc(votes)),
    pvote=votes/sum(votes)
  )

ggplot(res_cat, aes(x=rank, y=100*pvote)) +
  geom_bar(stat="identity", fill="grey50") +
  facet_grid(cat ~ .) +
  geom_text(
    aes(label=gsub(".* ([A-Za-z]+)$", "\\1", name)),
    y=0.3,
    angle=90,
    hjust=0
  )+
  theme_sixtysix() +
  geom_vline(xintercept=8.5, linetype="dashed") +
  labs(x="Rank", y="% of Vote", title="Results by Division Bloc")

The top eight candidates in the Wealthy Progressive divisions received on average 9.1% of the vote, compared to 8.2% in the Black Voter divisions, 7.9% in the Hispanic Voter, and 7.5% in the White Moderate. To measure another way, Common Pleas candidates had a gini coefficient of votes of 0.29 in Wealthy Progressive divisions, compared to 0.22, 0.19, and 0.14 in the other three blocs, indicating greater inequality in the votes candidates received, and thus more separation.

This is probably due to voters there being more likely to look up recommendations either on or before election day, and having consolidated preferences.

View code
gini <- function(x){
  outer_sum <- outer(x, x, FUN="-")
  gini <- sum(abs(outer_sum)) / (2 * length(x)^2 * mean(x))
  return(gini)
}

# res_cat %>%
#   group_by(cat) %>%
#   summarise(
#     gini = gini(pvote),
#     mean=mean(pvote[rank <= 8])
#   )

Mail-In Votes

Entering this election, I was especially interested if mail-in ballots would have different dynamics than in-person voting. When people vote at the kitchen table, likely over days, will ballot position matter less? Will endorsements matter more?

In total, 33% of the votes for CP came by Mail, vs 66% on Election Day (and 1% Provisionals). The Wealthy Progressive divisions were more likely to use mail: 38% of their votes were by mail, compared to 34% in the White Moderate divisions, and 29 and 28% in the Black and Hispanic Voter divisions.

View code
# res_cp %>% group_by(vote_type) %>%
#   summarise(votes=sum(votes)) %>%
#   mutate(pct=votes/sum(votes))
# 
# res_cp %>% 
#   left_join(div_cats, by=c("division"="row_id")) %>%
#   group_by(cat, vote_type) %>%
#   summarise(votes=sum(votes)) %>%
#   group_by(cat) %>%
#   mutate(pct=votes/sum(votes))

The candidates who did better by mail were all in the bottom right of the ballot. The top three were also Highly Recommended, suggesting that endorsements were more likely to be looked up by people voting by mail.

View code
res_votetype <- res_cp %>%
  filter(!is.na(name)) %>%
  group_by(vote_type, name, rownumber, colnumber, philacommrec, dcc) %>%
  summarise(votes=sum(votes)) %>%
  group_by(vote_type) %>%
  mutate(pvote=votes/sum(votes)) %>%
  group_by(name) %>%
  mutate(votes_total=sum(votes)) %>%
  ungroup() %>%
  pivot_wider(names_from=vote_type, values_from=c(votes, pvote)) %>%
  mutate(pvote_total=votes_total / sum(votes_total)) %>%
  arrange(desc(pvote_Mail - `pvote_Election Day`))

ggplot(
  res_votetype %>% arrange(votes_total) %>% mutate(winner = rank(-votes_total) <= 8) %>%
    mutate(diff=pvote_Mail - `pvote_Election Day`),
  aes(y=rownumber, x=colnumber)
) +
  geom_tile(
    aes(fill=100*diff),
    size=2
  ) +
  geom_text(
    aes(
      label = ifelse(philacommrec==1, "R", ifelse(philacommrec==2,"HR","")),
      x=colnumber+0.45,
      y=rownumber+0.45
    ),
    color="grey70",
    hjust=1, vjust=0
  ) +
  geom_text(
    aes(
      label = ifelse(dcc==1, "D", ""),
      x=colnumber-0.45,
      y=rownumber+0.45
    ),
    color="grey70",
    hjust=0, vjust=0
  ) +
  geom_text(
    aes(
      label = sprintf("%s\n%s%0.1f%%, (%0.1f%% - %0.1f%%)", name, ifelse(diff>0, "+", "-"),100*abs(diff), 100*pvote_Mail,100* `pvote_Election Day`)
    ),
    color="black"
    # fontface="bold"
  ) +
  scale_y_reverse(NULL) +
  scale_x_continuous(NULL)+
  scale_fill_viridis_c(guide=FALSE) +
  annotate(
    "text",
    label="R = Recommended\nHR = Highly Recommended\nD = DCC Endorsed",
    x = 1.6,
    y = 6,
    hjust=0,
    vjust=0.5,
    color="grey70"
  ) +
  theme_sixtysix() %+replace% 
  theme(
    panel.grid.major=element_blank(),
    axis.text=element_blank()
  ) +
  ggtitle(
    "Common Pleas: Mail minus In Person.",
    "2021 Democratic Primary, arranged as on the ballot."
  )

Some of these differences are due to differential rates of mail-in usage: Moore, Hangley, and Wahl all did better in Wealthy Progressive wards, and they mailed in more often. We can adjust that by taking the within-Division difference for each candidate, and then taking a weighted average across divisions weighted by total votes. This decomposition leads to basically the same finding but slightly smaller effects: Moore did 1.9 percentage points better by mail within a given division, Hangley 1.2, and Hall 0.9.

Some of this may still be because the mail voters were systematically different from in-person voters, but this within-division comparison is the closest we can get with the available data.

Doing the decomposition by Voting Bloc is interesting. Moore does better everywhere by mail than in person, probably reflecting the important DCC and Bar effects but a lack of endorsements with Election Day presence. In every bloc, the highest differences are the poor-ballot-position, Bar-recommended candidates. Interestingly, the largest differences are in the White Moderate divisions (South Philly and the Northeast), probably reflecting the politicized nature of mail-in voting, and that in those divisions there were political differences between those who voted by mail and in person.

View code
res_votetype_weighted <- res_cp %>%
  select(division, name, vote_type, votes) %>%
  filter(!is.na(name)) %>%
  group_by(division, vote_type) %>%
  mutate(pvote=votes/sum(votes)) %>%
  group_by(division) %>%
  mutate(total_votes=sum(votes)) %>%
  select(-votes) %>%
  pivot_wider(names_from=vote_type, values_from=pvote) %>%
  mutate(diff = `Mail` - `Election Day`) %>%
  group_by(name) %>%
  summarise(
    diff=weighted.mean(diff, w=total_votes, na.rm = T)
  )

res_votetype_weighted_cat <- res_cp %>%
  select(division, name, vote_type, votes) %>%
  filter(!is.na(name)) %>%
  group_by(division, vote_type) %>%
  mutate(pvote=votes/sum(votes)) %>%
  group_by(division) %>%
  mutate(total_votes=sum(votes)) %>%
  select(-votes) %>%
  left_join(div_cats, by=c("division"="row_id")) %>%
  pivot_wider(names_from=vote_type, values_from=pvote) %>%
  mutate(diff = `Mail` - `Election Day`) %>%
  group_by(name, cat) %>%
  summarise(
    diff=weighted.mean(diff, w=total_votes, na.rm = T)
  )

ggplot(
  res_votetype_weighted_cat %>% group_by(cat) %>%
    mutate(rank=rank(desc(diff))),
  aes(x=rank, y=100*diff)
) +
  geom_bar(stat="identity") +
  facet_grid(cat~.) +
  theme_sixtysix() +
  geom_text(aes(label= gsub(".* ([A-Za-z]+)$", "\\1", name)), angle=90, hjust=0, y=0.1)+
  labs(
    title="Mail minus In-Person Results", subtitle="By Voting Bloc",
    y="Mail % minus In-Person %",
    x=NULL
  )

Next: The Effect of the Bar Recommendations!

Coming soon.