#### Paper title: Polls of fear? Electoral violence, incumbent strength, and voter turnout in Côte d'Ivoire
#### Purpose: Descriptive analysis
#### Author: Sebastian van Baalen
#### Date: 2023-01-04
#### Read this first ####
# If the script contains errors or fails to run, compare the package version
# with the version listed in the run-file.
#### Front matter ####
# Install packages
# Un-comment to install packages
#install.packages("here") # For using relative paths
#install.packages("ggplot2") # For making graphs and figures
#install.packages("tidyverse") # For working with tidy data
#install.packages("sf") # For working with spatial data
#install.packages("viridis") # For using the Viridis color scales
# Load packages
library(here) # For using relative paths
## here() starts at /Users/sebva935/Documents/Datasets and scripts/Replication files/JPR 2022
library(ggplot2) # For making graphs and figures
library(tidyverse) # For working with tidy data
## ── Attaching packages
## ───────────────────────────────────────
## tidyverse 1.3.2 ──
## ✔ tibble 3.1.8 ✔ dplyr 1.0.10
## ✔ tidyr 1.2.1 ✔ stringr 1.4.1
## ✔ readr 2.1.3 ✔ forcats 0.5.2
## ✔ purrr 0.3.5
## ── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()
library(sf) # For working with spatial data
## Linking to GEOS 3.10.2, GDAL 3.4.2, PROJ 8.2.1; sf_use_s2() is TRUE
library(viridis) # For using the Viridis color scales
## Loading required package: viridisLite
# Load electoral violence dataset
load("ev.Rdata")
# Set theme
my_theme <-
theme_bw() +
theme(
axis.title.y = element_text(margin = margin (t = 0, r = 10, b = 0, l = 0), size = 16),
axis.title.x = element_text(margin = margin (t = 10, r = 0, b = 0, l = 0), size = 16),
axis.text.y = element_text(size = 14),
axis.text.x = element_text(size = 12)
)
#### Data limitations and reporting ####
# Create source variables
ev <- ev %>%
mutate(
MultipleSources = case_when(
grepl(";", Source) ~ "Multiple.Sources",
TRUE ~ "Single.Source"
),
ACLED = case_when(
grepl("ACLED", Source) ~ 1,
TRUE ~ 0
),
CNDH = case_when(
grepl("CNDH", Source) ~ 1,
TRUE ~ 0
),
ACLEDCNDH = case_when(
grepl("ACLED", Source) ~ 1,
grepl("CNDH", Source) ~ 1,
TRUE ~ 0
),
) %>%
separate(col = EventName, into = c("EventName1", "EventName2"), sep = ";", remove = FALSE)
## Warning: Expected 2 pieces. Missing pieces filled with `NA` in 301 rows [1, 2,
## 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, ...].
##### Number and share of events reported by multiple sources ####
ev %>%
group_by(MultipleSources) %>%
summarise(Frequency = n()) %>%
ungroup %>%
mutate('Share of events (%)' = Frequency/sum(Frequency) * 100)
## # A tibble: 2 × 3
## MultipleSources Frequency `Share of events (%)`
## <chr> <int> <dbl>
## 1 Multiple.Sources 49 15.3
## 2 Single.Source 271 84.7
##### Number and share of events reported by multiple sources by event type ####
ev %>%
group_by(EventName1, MultipleSources) %>%
summarise(Frequency = n()) %>%
pivot_wider(names_from = MultipleSources, values_from = Frequency) %>%
ungroup %>%
mutate_if(is.numeric, ~replace_na(., 0)) %>%
mutate('Share of events (%)' = Multiple.Sources/(Multiple.Sources + Single.Source) * 100)
## `summarise()` has grouped output by 'EventName1'. You can override using the
## `.groups` argument.
## # A tibble: 16 × 4
## EventName1 Multiple.Sources Single.Source Share of e…¹
## <chr> <int> <int> <dbl>
## 1 Arrest 1 4 20
## 2 Arson 0 20 0
## 3 Attack 8 20 28.6
## 4 Clash 24 38 38.7
## 5 Intimidation 0 11 0
## 6 Kidnapping 0 1 0
## 7 Killing 0 1 0
## 8 Looting 1 56 1.75
## 9 Protest 1 0 100
## 10 Protest with intervention 2 6 25
## 11 Riot 0 1 0
## 12 Roadblocks 2 26 7.14
## 13 Shooting 0 1 0
## 14 Unrest 0 10 0
## 15 Violent protest 2 50 3.85
## 16 Violent protest with intervention 8 26 23.5
## # … with abbreviated variable name ¹​`Share of events (%)`
##### Number and share of events reported by multiple sources by bodily injury ####
ev %>%
mutate(
Casualties = case_when(
ParticipantDeathsHigh > 0 ~ "Casualties",
ParticipantInjuriesHigh > 0 ~ "Casualties",
TRUE ~ "No.Casualties"
)
) %>%
group_by(Casualties, MultipleSources) %>%
summarise(Frequency = n()) %>%
pivot_wider(names_from = MultipleSources, values_from = Frequency) %>%
ungroup %>%
mutate_if(is.numeric, ~replace_na(., 0)) %>%
mutate('Share of events (%)' = Multiple.Sources/(Multiple.Sources + Single.Source) * 100)
## `summarise()` has grouped output by 'Casualties'. You can override using the
## `.groups` argument.
## # A tibble: 2 × 4
## Casualties Multiple.Sources Single.Source `Share of events (%)`
## <chr> <int> <int> <dbl>
## 1 Casualties 30 51 37.0
## 2 No.Casualties 19 220 7.95
##### Number of events by timing ####
ev %>%
mutate(
Date = as.Date(Date, format = "%Y-%m-%d"),
Timing = case_when(
Date < "2020-10-31" ~ "Pre-election",
Date == "2020-10-31" ~ "Presidential election day",
Date > "2020-10-31" & Date < "2021-03-06" ~ "Interim period",
Date == "2021-03-06" ~ "Legislative election day"
)
) %>%
group_by(Timing) %>%
summarise(Total = n())
## # A tibble: 4 × 2
## Timing Total
## <chr> <int>
## 1 Interim period 53
## 2 Legislative election day 14
## 3 Pre-election 155
## 4 Presidential election day 98
#### Visualizations ####
##### Bar graph of number of events by type ####
# Note that this graph focuses on how (through what type of action) the event started
ev %>%
ggplot() +
geom_bar(aes(x = EventName1), fill = "grey68") +
scale_y_continuous(breaks = seq(0, 60, 10)) +
labs(x = NULL, y = "Number of events") +
my_theme +
theme(axis.text.x = element_text(hjust = 1, size = 12, angle = 45))
ggsave("event_type.png", width = 14, height = 8.9)
# Number of deaths and injuries by event type
ev %>%
group_by(EventName1) %>%
summarise(
Deaths = sum(ParticipantDeaths),
Injured = sum(ParticipantInjuries)
) %>%
select(EventName1, Deaths, Injured)
## # A tibble: 16 × 3
## EventName1 Deaths Injured
## <chr> <int> <int>
## 1 Arrest 0 0
## 2 Arson 0 0
## 3 Attack 12 34
## 4 Clash 71 583
## 5 Intimidation 0 3
## 6 Kidnapping 0 0
## 7 Killing 1 0
## 8 Looting 0 1
## 9 Protest 0 3
## 10 Protest with intervention 3 5
## 11 Riot 0 0
## 12 Roadblocks 0 3
## 13 Shooting 0 0
## 14 Unrest 0 0
## 15 Violent protest 0 3
## 16 Violent protest with intervention 4 47
# Total number of deaths and injuries
ev %>% summarise(Deaths = sum(ParticipantDeaths), Injured = sum(ParticipantInjuries))
## Deaths Injured
## 1 91 682
##### Bar graph of the number of casualties by type of violence ####
ev %>%
separate(col = EventName, into = c("EventName1", "EventName2"), sep = ";") %>%
group_by(EventName1) %>%
summarise(
Casualties = sum(ParticipantDeaths, ParticipantInjuries),
CasualtiesHigh = sum(ParticipantDeathsHigh, ParticipantInjuriesHigh),
CasualtiesLow = sum(ParticipantDeathsLow, ParticipantInjuriesLow)
) %>%
ggplot() +
geom_bar(aes(x = EventName1, y = Casualties), stat = "identity", fill = "grey68") +
geom_errorbar(aes(x = EventName1, ymin = CasualtiesLow, ymax = CasualtiesHigh), stat = "identity", linewidth = .8, width = 0.2) +
scale_y_continuous(breaks = seq(0, 700, 100)) +
labs(x = NULL, y = "Number of casualties") +
my_theme +
theme(axis.text.x = element_text(hjust = 1, size = 12, angle = 45))
## Warning: Expected 2 pieces. Missing pieces filled with `NA` in 301 rows [1, 2,
## 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, ...].
ggsave("event_type_casualties.png", width = 14, height = 8.9)
##### Bar graph of share of events by involvement ####
positions <- c("Not reported", "Only opposition", "Only government", "Both opposition and government")
ev %>%
mutate(
Involvement = case_when(
Actor1Side == 1 & Actor2Side == -99 ~ "Only opposition",
Actor1Side == 1 & is.na(Actor2Side) ~ "Only opposition",
Actor1Side == 1 & Actor2Side == 1 ~ "Only opposition",
Actor1Side == -99 & Actor2Side == 0 ~ "Only government",
Actor1Side == 0 & Actor2Side == 0 ~ "Only government",
Actor1Side == 0 & is.na(Actor2Side) ~ "Only government",
Actor1Side == 1 & Actor2Side == 0 ~ "Both opposition and government",
Actor1Side == 0 & Actor2Side == 1 ~ "Both opposition and government",
TRUE ~ "Not reported"
)
) %>%
select(Actor1Side, Actor2Side, Involvement) %>%
ggplot() +
geom_bar(aes(y = (after_stat(count))/sum(after_stat(count)), x = Involvement), fill = "grey68") +
scale_y_continuous(labels = scales::percent_format(accuracy = 5L), breaks = seq(0, 0.5, 0.1), limits = c(0, 0.5)) +
scale_x_discrete(limits = positions) +
labs(x = NULL, y = "Share of events") +
my_theme
ggsave("involvement.png", width = 12, height = 8)
##### Bar graph of the share of events by initiator ####
ev <- ev %>%
mutate(
Initiator = case_when(
Actor1Side == 1 & EventDirection == 0 ~ "Opposition-initiated",
Actor1Side == 1 & ViolenceInitiator == "0,1" ~ "Opposition-initiated",
Actor1Side == 0 & EventDirection == 0 ~ "Government-initiated",
Actor1Side == 0 & ViolenceInitiator == "0,1" ~ "Government-initiated",
Actor2Side == 0 & ViolenceInitiator == "0,2" ~ "Government-initiated",
TRUE ~ "Initiator not reported"
)
)
table(ev$Actor1Name[ev$Initiator == "Initiator not reported"])
##
## A group of men
## 1
## Ethnic community supporting the opposition (PDCI)
## 1
## Government supporters
## 1
## Machette-armed man
## 1
## Microbes
## 1
## Opposition supporters
## 2
## Opposition supporters (assumed)
## 8
## Opposition supporters (FPI-Laurent Gbagbo)
## 1
## Opposition supporters (FPI)
## 1
## Opposition supporters (GPS, FPI, PDCI)
## 1
## Residents
## 1
## Unidentified armed individuals
## 11
## Unidentified armed youths
## 1
## Unidentified individuals
## 94
## Unidentified men
## 14
## Unidentified youth
## 9
# Order categories in chart
positions <- c("Opposition-initiated", "Government-initiated", "Initiator not reported")
ev %>%
ggplot() +
geom_bar(aes(y = (..count..)/sum(..count..), x = Initiator), fill = "grey68") +
scale_x_discrete(limits = positions) +
scale_y_continuous(labels = scales::percent_format(accuracy = 5L), breaks = seq(0, 0.5, 0.1), limits = c(0, 0.5)) +
labs(x = NULL, y = "Share of events") +
my_theme
## Warning: The dot-dot notation (`..count..`) was deprecated in ggplot2 3.4.0.
## ℹ Please use `after_stat(count)` instead.
ggsave("initiator.png", width = 12, height = 8)
##### Bar graph of the number of events by type and timing ####
ev %>%
mutate(
Date = as.Date(Date, format = "%Y-%m-%d"),
Timing = case_when(
Date < "2020-10-31" ~ "Pre-election",
Date == "2020-10-31" ~ "Presidential election day",
Date > "2020-10-31" & Date < "2021-03-06" ~ "Interim period",
Date == "2021-03-06" ~ "Legislative election day"
)
) %>%
ggplot() +
geom_bar(aes(x = EventName1), fill = "grey68") +
facet_wrap(~ factor(Timing, levels = c("Pre-election", "Presidential election day", "Interim period", "Legislative election day"))) +
labs(x = NULL, y = "Number of events") +
my_theme +
theme(
axis.text.x = element_text(hjust = 1, size = 12, angle = 45),
strip.text.x = element_text(size = 14)
)
ggsave("type_and_timing.png", width = 16, height = 12)
##### Number of electoral violence events by voting district ####
load("df.Rdata")
# Load shapefile
shapefile <- st_read("shapefile.shp")
## Reading layer `shapefile' from data source
## `/Users/sebva935/Documents/Datasets and scripts/Replication files/JPR 2022/shapefile.shp'
## using driver `ESRI Shapefile'
## Simple feature collection with 205 features and 3 fields
## Geometry type: MULTIPOLYGON
## Dimension: XY
## Bounding box: xmin: -8.584147 ymin: 4.360553 xmax: -2.492309 ymax: 10.74565
## Geodetic CRS: WGS 84
# Merge electoral results and shapefile
shapefile <- shapefile %>% rename(Vote.District = Legislativ)
df <- shapefile %>%
left_join(df, by = "Vote.District")
# Make electoral violence dataset spatial to enable geo-processing
ev <- as.data.frame(ev) %>%
st_as_sf(coords = c("Longitude", "Latitude"), crs = 4326, remove = FALSE)
# Create electoral violence variable
df <- df %>%
mutate(
Electoral.Violence.Events = lengths(st_intersects(x = df, y = ev))
)
# Make map
df %>%
ggplot() +
geom_sf(
aes(
fill = Electoral.Violence.Events
),
color = "white",
linewidth = 0.1,
) +
coord_sf(
xlim = c(-9, -2),
ylim = c(4, 11),
expand = FALSE
) +
scale_fill_viridis(
name = "Number of events",
option = "viridis",
discrete = FALSE,
begin = 0,
end = 1,
alpha = 0.8,
direction = 1,
na.value = "white"
) +
annotate(
"text",
x = -3.55,
y = 9.1,
label = "Bouna\nNational\nPark",
size = 3.5,
lineheight = 1,
hjust = "center"
) +
theme_void()
ggsave("map.png", width = 10, height = 8.9)
##### Number of electoral violence events by voting district in Abidjan ####
df %>%
mutate(Sub.Prefecture = toupper(Sub.Prefecture)) %>%
filter(Region == "District Autonome D'Abidjan") %>%
mutate(
Sub.Prefecture1 = case_when(
Sub.Prefecture != "MARCORY" ~ Sub.Prefecture
),
Sub.Prefecture2 = case_when(
Sub.Prefecture == "MARCORY" ~ "MARCORY"
)
) %>%
ggplot() +
geom_sf(
aes(
fill = Electoral.Violence.Events
),
color = "white",
size = 0.1,
) +
coord_sf(
xlim = c(-4.35, -3.75),
ylim = c(5.2, 5.55),
expand = FALSE
) +
scale_fill_viridis(
name = "Number of events",
option = "viridis",
discrete = FALSE,
begin = 0,
end = 1,
alpha = 0.8,
direction = 1,
na.value = "white"
) +
geom_sf_label(aes(label = Sub.Prefecture1), size = 3) +
geom_sf_label(aes(label = Sub.Prefecture2), size = 3, nudge_y = 0.013) +
theme_void()
## Warning in st_point_on_surface.sfc(sf::st_zm(x)): st_point_on_surface may not
## give correct results for longitude/latitude data
## Warning in st_point_on_surface.sfc(sf::st_zm(x)): st_point_on_surface may not
## give correct results for longitude/latitude data
## Warning: Removed 1 rows containing missing values (`geom_label()`).
## Warning: Removed 12 rows containing missing values (`geom_label()`).
ggsave("map_abidjan.png", width = 10, height = 6)
## Warning in st_point_on_surface.sfc(sf::st_zm(x)): st_point_on_surface may not
## give correct results for longitude/latitude data
## Warning in st_point_on_surface.sfc(sf::st_zm(x)): st_point_on_surface may not
## give correct results for longitude/latitude data
## Warning: Removed 1 rows containing missing values (`geom_label()`).
## Warning: Removed 12 rows containing missing values (`geom_label()`).
##### Bar graph of the number of events by strongholds (2010) ####
df %>%
ggplot() +
geom_bar(aes(x = Winner.Name.2010, y = Electoral.Violence.Events/sum(Electoral.Violence.Events)), stat = "identity", fill = "grey68") +
scale_y_continuous(labels = scales::percent_format(accuracy = 5L), breaks = seq(0, 1, 0.1), limits = c(0, 1)) +
labs(x = NULL, y = "Share of events") +
my_theme
ggsave("strongholds2010.png", width = 12, height = 8)
##### Bar graph of the number of events by strongholds (2021) ####
df %>%
mutate(
Stronghold2021 = case_when(
Winner.Name.2021 == "RHDP" ~ "Incumbent stronghold (RHDP)",
Winner.Name.2021 == "Independent" ~ "Independents",
TRUE ~ "Opposition stronghold"
),
) %>%
group_by(Stronghold2021) %>%
summarise(Events = sum(Electoral.Violence.Events)) %>%
ggplot() +
geom_bar(aes(x = Stronghold2021, y = Events/sum(Events)), stat = "identity", fill = "grey68") +
scale_y_continuous(labels = scales::percent_format(accuracy = 5L), breaks = seq(0, 1, 0.1), limits = c(0, 1)) +
labs(x = NULL, y = "Share of events") +
my_theme
ggsave("strongholds2021.png", width = 12, height = 8)