R code to produce figure for "How men can help in the fight for women’s rights" at the WaPo's Monkey Cage

monkeycage_figure.R

################################################################################
# Alluvial Plot of Abortion Attitudes for The Washington Post's Monkey Cage
# Data: 2010-2014 Cooperative Congressional Election Study Panel Survey
################################################################################


# ---- Setup ----
# install.packages("tidyverse")
# install.packages("stringr")
# install.packages("haven")
# install.packages("easyalluvial")
# install.packages("wesanderson")

library(tidyverse)
library(stringr)
library(haven)
library(easyalluvial)
library(wesanderson)


# ---- Data ----
# https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/TOE8I1&version=11.0

df_cces <- haven::read_dta("https://dataverse.harvard.edu/api/access/datafile/:persistentId?persistentId=doi:10.7910/DVN/TOE8I1/90WWPS")


# ---- Recodes ----

df_cces %>%
  mutate(
    abort_2010 = case_when(
      CC10_324 == 1 ~ "Never legal",
      CC10_324 == 2 ~ "Rape, incest,\nlife of mother",
      CC10_324 == 3 ~ "Permitted with\nsome limits",
      CC10_324 == 4 ~ "Always legal",
      T ~ NA_character_
    )%>% factor(., levels = c("Never legal", 
                              "Rape, incest,\nlife of mother",
                              "Permitted with\nsome limits", 
                              "Always legal")
                ),
    abort_2012 = case_when(
      CC12_324 == 1 ~ "Never legal",
      CC12_324 == 2 ~ "Rape, incest,\nlife of mother",
      CC12_324 == 3 ~ "Permitted with\nsome limits",
      CC12_324 == 4 ~ "Always legal",
      T ~ NA_character_
    )%>% factor(., levels = c("Never legal", 
                              "Rape, incest,\nlife of mother",
                              "Permitted with\nsome limits", 
                              "Always legal")
                ),
    abort_2014 = case_when(
      CC14_324 == 1 ~ "Never legal",
      CC14_324 == 2 ~ "Rape, incest,\nlife of mother",
      CC14_324 == 3 ~ "Permitted with\nsome limits",
      CC14_324 == 4 ~ "Always legal",
      T ~ NA_character_
    ) %>% factor(., levels = c("Never legal", 
                               "Rape, incest,\nlife of mother",
                               "Permitted with\nsome limits", 
                               "Always legal")
                 )
  ) -> df_cces

# ---- Calculate percentages by year ----

df_cces %>%
  select(starts_with("abort")) %>%
  pivot_longer(
    cols = starts_with("abort_")
  )%>%
  mutate(
    x = gsub("abort_", "", name)
  ) %>%
  group_by(x) %>%
  summarise(
    "Always legal" = mean(value == "Always legal", na.rm=T),
    "Permitted with\nsome limits" = mean( value == "Permitted with\nsome limits", na.rm=T),
    "Rape, incest,\nlife of mother" = mean(value == "Rape, incest,\nlife of mother", na.rm=T),
    "Never legal" = mean(value == "Never legal",na.rm=T)
  ) %>%
  pivot_longer(
    cols = -x,
    names_to = "label",
    values_to = "percent"
  ) %>%
  mutate(
    label = factor(label, 
                   levels = c("Never legal", 
                              "Rape, incest,\nlife of mother",
                              "Permitted with\nsome limits", 
                              "Always legal")
                   ),
    percent = scales::percent(percent, accuracy = 0.1)
  ) -> df_percent

## Double check proportions

# df_percent
# prop.table(table(df_cces$abort_2010))
# prop.table(table(df_cces$abort_2012))
# prop.table(table(df_cces$abort_2014))

# ---- Create figure ----

# Custom color palette from mid Wes Anderson Film
col_vector = wes_palette(n=4, name="GrandBudapest2")[c(1,3,2,4)]


## Baseline figure

df_cces %>%
  select(starts_with("abort_")) %>%
  na.omit()%>%
  rename_all(~sub("abort_*", "", .))%>%
  alluvial_wide(fill_by = 'first_variable',
                verbose = F,
                col_vector_flow = col_vector,
                col_vector_value = col_vector,
                stratum_labels = F
  )+
  labs(
    y = "",
    title = "Which opinion best agrees with your view on abortion?"
  )+
  theme_minimal()+
  scale_x_discrete(expand = expansion(add = c(.75,.25))
                   )+
  theme(axis.text.y=element_blank(),
        axis.ticks.y=element_blank()
        ) -> p

## Update caption
p$labels$caption <- "Data: 2010-2014 Cooperative Congressional Election Study Panel Survey\nFigure: Tarah Williams and Paul Testa" 

## Merge in percents
p$data <- p$data %>% left_join(df_percent, by = c("x" = "x", "value" = "label")
                               )

## Add percentage labels labels
p <- p+ geom_label(stat = ggalluvial::StatStratum, aes(label = percent, group = fill_value))

## Annotate y axis

## Axis Labels
the_labs <- levels(df_cces$abort_2014)

## Axis heights
p$data%>%
  filter(x == 2010)%>%
  group_by(fill)%>%
  summarise(
    n = sum(n),
  )%>%
  ungroup()%>%
  arrange(desc(fill))%>%
  mutate(
   height = n/2 + lag(cumsum(n))
  )

the_heights <- c(478,2180, 4088, 7046)

p + annotate("text", x = 0.25, 
             y= the_heights, 
             label = the_labs, 
             hjust = 0,
             size = 4)+
  theme(panel.grid.major = element_blank(), 
        panel.grid.minor = element_blank()
        ) -> p

p

## Save image
# ggsave("cces_abortion_alluvial.jpeg", 
#        plot = p,
#        bg = "white",
#        width = 8,
#        height = 5
#       )
#
# ggsave("cces_abortion_alluvial.png",
#        plot = p,
#        bg = "white",
#        width = 8,
#        height = 5
#       )