The world is progressively embracing democracy

A reproduction of the Our World in Data visualization, which is based on the classification and assessment of countries’ political regimes by Skaaning et al. (2015).

Democracy

The graphic I choose shows how democracy has spread across countries, how it differs between them, and whether we are moving towards a more democratic world. The variables denoted in the chart shows the number of electoral democracies in the world based on the Lexical Index of Electoral Democracy-classification from Skaaning et al. (2015).

The Lexical Index definition attained from Our World in Data establishes that it is a classification system that categorizes political systems or regimes based on easily observable characteristics. It aims to avoid difficult evaluations by experts or researchers and relies on its own teams to assess specific criteria. These criterions are used to identify democracies and non-democracies. The index is able to track political systems over time, measured since the year 1789, and includes 242 countries. The index prioritizes straightforward and objective factors, making it a tool for quick assessments and comparisons.

Definitions

Types of Democracy in the Lexical Index:

1. Non-Electoral Autocracy: A political system characterized by a single entity or a small group holding power without elections, often with a single autocrat or ruling elite.

2. One-Party Autocracy: A system where a single political party maintains absolute control over the government, suppressing the opposition and competitive elections.

3. Multi-Party Autocracy without Elected Executive: A political system with multiple parties, but the executive leader is not elected democratically.

4. Multi-Party Autocracy: Features multiple political parties, but the political system still retains autocratic elements.

5. Exclusive Democracy: Inclusive of elections but excludes certain groups or individuals from political participation.

6. Male Democracy: A democracy where only male citizens are allowed to participate in elections.

7. Electoral Democracy: A political system with regular elections where citizens can choose their leaders, but this may not guarantee full democratic rights.

8. Polyarchy: A comprehensive democracy with competitive elections, civil liberties, political pluralism, and broad participation.

These definitions reflect the categorization of political systems within the Lexical Index, based on specific characteristics and levels of inclusiveness attained from Skaaning et al. (2015), with “Polyarchy” representing the highest level of democracy and inclusivity.

Getting the data

The dataset, obtained from Our World in Data and based on research by Skaaning et al. (2015), categorizes the world’s nations into democracies and autocracies. This graphic dynamically captures the shifts in these counts from 1789 through 2022, offering insights into the evolving landscape of global political systems nowadays. All the data needed for this chart can be downloaded directly from the Our World in Data website in the topic Democracy (https://ourworldindata.org/grapher/countries-democracies-autocracies-lexical).

data <- read.csv("data_world.csv", sep = ";", dec = ",")

Building the chart

library(ggplot2)
library(dplyr)
library(tidyr)

Data

The data scoured from Our World in Data needed to be converted to frequency tables . The graph’s vertical axis, which I aimed to replicate, utilized cumulative frequency as a metric, necessitating the data’s conversion. Subsequently, I identified and chose the specific columns needed for the replication, resulting in the creation of a new data-set named df_reduced.

df_reduced <- data %>%
  select(Year, Frec.Non.electoral.autocracies, Frec.One.party.autocracies,
         Frec.Multi.party.autocracies.without.elected.executive,
         Frec.Multi.party.autocracy, Frec.Exclusive.democracies, 
         Frec.Male.democracies, Frec.Electoral.democracies, 
         Frec.Polyarchies) 

custom_order <- c("Frec.Non.electoral.autocracies", "Frec.One.party.autocracies", 
                  "Frec.Multi.party.autocracies.without.elected.executive",
                  "Frec.Multi.party.autocracy", "Frec.Exclusive.democracies",
                  "Frec.Male.democracies", "Frec.Electoral.democracies",
                  "Frec.Polyarchies")

To enhance accessibility and facilitate the chart-building process, I converted the new dataset into tidy data. During this conversion, several changes were implemented, and intervals were established for the axis.

df_long <- df_reduced %>%
  pivot_longer(cols = -Year, names_to = "Democracy_Type", values_to = "n")

df_long <- df_long %>%
  mutate(Democracy_Type = factor(Democracy_Type, levels = custom_order))

interval_x <- c(1789, 1850, 1900, 1950, 2000, 2022)

interval_y <- seq(0, 100, by = 0.2)   

Due to each variable having an associated color. The following color vector was created for each value.

democracy_colors <- c(
  "Frec.Non.electoral.autocracies" = "#d73027",
  "Frec.One.party.autocracies" = "#f46d43", 
  "Frec.Multi.party.autocracies.without.elected.executive" = "#fdae61", 
  "Frec.Multi.party.autocracy" = "#fee090", 
  "Frec.Exclusive.democracies" = "#e0f3f8",
  "Frec.Male.democracies" = "#abd9e9",
  "Frec.Electoral.democracies" = "#74add1", 
  "Frec.Polyarchies" = "#4575b4"
)

Starting with the graphic

To construct the visual representation, I utilized the geom_area function from ggplot, tailored for generating Stacked Area Charts.

plot <- ggplot(df_long, aes(x = Year, y = n, fill = Democracy_Type)) +
  geom_area(alpha = 0.8, size = 0.2, aes(color = Democracy_Type)) +
  labs(x = NULL, y = NULL, fill = NULL) +
  scale_x_continuous(limits = c(1789, 2085), breaks = interval_x, minor_breaks = NULL) +
  scale_y_continuous(labels = scales::percent_format(scale = 100), breaks = interval_y) +
  scale_fill_manual(values = democracy_colors) +
  scale_color_manual(values = democracy_colors) +
  coord_cartesian(expand=0) +
  theme_minimal() +
  theme(
    axis.text.x = element_text(
      color="#5b5b5b", size = 10, family = "Lato", angle = 0, 
      hjust = c(0.1, 0.3, 0.5, 0.5, 0.5, 1), margin = margin(t = 0)),
    axis.ticks.x = element_line(color="#5b5b5b", size = 0.2),
    axis.ticks.length = unit(0.1, "cm"),
    legend.position = "none",
    panel.grid = element_blank(),
    axis.text.y = element_text(
      color="#5b5b5b", size = 10, family = "Lato", margin = margin(r = 0))
  ) +
  ggtitle(
    "Countries that are democracies and autocracies, World", 
    subtitle = "Political regimes based on the classification and assessment by Skaaning et al. (2015).") +
  theme(plot.title = element_text(
    color="#5b5b5b", family = "serif", size = 15, face = "bold", hjust = -0.27),
        plot.subtitle = element_text(
          color="#5b5b5b", family = "Lato", size = 9, hjust = -0.24)
  )
plot

I employed the intervals created during the data organization for both the x-axis and the y-axis in constructing the graphic. Given that the y-axis represents frequencies in percentage, the code was modified to reflect this in the visualization. The year limits were adjusted to fit the dimensions required for displaying custom labels for each democratic type. While the code’s limits span from 1789 to 2085, the graphic displays the period from 1789 to 2022 on the x-axis.

To achieve a custom separation between each year on the x-axis, adjustments were made in the theme section, specifically in axis.text.x, where the spacing between each year was manually configured in order to be exactly the distance displayed in the graphic of reference.

Final touches

I generated custom labels with corresponding colors and segments, mirroring the structure observed in the graphic that I aimed to replicate.

custom_legend <- data.frame(
  Democracy_Type = levels(df_long$Democracy_Type),
  x = rep(2022, length(levels(df_long$Democracy_Type))),
  y = seq(0, 1, length.out = length(levels(df_long$Democracy_Type))),
  label = c("Polyarchies",
            "Electoral\ndemocracies",
            "Male\ndemocracies",
            "Exclusive\ndemocracies",
            "Multi-party\nautocracy",
            "Multi-party\nautocracies\nwithout elected\nexecutive",
            "One-party\nautocracies",
            "Non-electoral\nautocracies"
  ),
  label_distance = c(11, 11, 11, 21, 4, 31, 4, 8),
  segment_distance = c(10, 10, 10, 20, 3, 30, 3, 7), 
  vertical_offset_segment = c(0.20, 0.33, 0.338, -0.16, 0.15, 0.09, 0.01, -0.04),
  vertical_offset_label = c(0.20, 0.33, 0.338, -0.16, 0.15, 0.09, 0.01, -0.04)
)

corrected_Democracy_Type <- rev(custom_legend$Democracy_Type)

To replicate the specific placement of labels and segments in the target graphic, I modified the code to permit manual adjustments for the angle, vertical, and horizontal placement of both text and segments. Thus, segment_distance makes reference to the horizontal placement of the segment and vertical_offset_segment is for the vertical placement.

Furthermore, I introduced a custom legend to the plot and fine-tuned the size parameters to precisely match the dimensions of the desired graphic.

plot2 <- plot +
  geom_text(
    data = custom_legend,
    aes(x = x + label_distance, 
        y = y + vertical_offset_label, 
        label = label, 
        color = corrected_Democracy_Type),
    size = 2.5,
    family = "Lato",
    hjust = 0
  ) +
  geom_segment(
    data = custom_legend,
    aes(x = x + segment_distance, 
        xend = x + 1, 
        y = y + vertical_offset_segment, 
        yend = y + vertical_offset_segment, 
        color = "#5b5b5b"),
    size = 0.5
  ) +
  theme(
    plot.margin = margin(0.3, 0.3, 0.3, 0.3, "cm")
  )

plot2

Improving the chart

Upon careful consideration, I determined that there was no apparent need for enhancements in the replicated graphic. Consequently, I opted to generate a new visual representation utilizing the same dataset.

Building the new chart

In an effort to retain as much of the original code as possible, I generated a new custom color palette to address perceived deficiencies in the previous one. This adjustment was made to enhance visibility, ensuring that all colors are easily discernible. Additionally, I maintained the code for creating custom labels in the process as I wanted to highlight that code and expand the use of it to improve the new graphic.

democracy_colors <- c(
  "Frec.Non.electoral.autocracies" = "#081d58",
  "Frec.One.party.autocracies" = "#253494",
  "Frec.Multi.party.autocracies.without.elected.executive" = "#225ea8",
  "Frec.Multi.party.autocracy" = "#41b6c4",
  "Frec.Exclusive.democracies" = "#7fcdbb",
  "Frec.Male.democracies" = "#c7e9b4",
  "Frec.Electoral.democracies" = "#1d91c0",
  "Frec.Polyarchies" = "black"
)
custom_legend <- data.frame(
  Democracy_Type = levels(df_long$Democracy_Type),
  x = rep(2022, length(levels(df_long$Democracy_Type))),
  y = seq(0, 1, length.out = length(levels(df_long$Democracy_Type))),
  label = c(
    "Polyarchies",
    "Electoral democracies",
    "Male democracies","Exclusive democracies",
    "Multi-party autocracy",
    "Multi-party autocracies without elected executive",
    "One-party autocracies",
    "Non-electoral autocracies"
  ),
  label_distance = c(3, -20, 3, 3, -60, -220  , 3, -220),
  vertical_offset_label = c(0.38, 0.17,-0.28, -0.27, -0.59, -0.32, -0.76, 0)
)

corrected_Democracy_Type <- rev(custom_legend$Democracy_Type)

The chosen type of chart for improvements was a Parallel Plot utilizing the geom_line function from ggplot. I chose this approach as it provides a clearer representation of the evolution of each democracy type throughout the analysis period.

In trying to create the new graphic, I aimed for a fresh look while retaining the essence of the original one I sought to replicate. To achieve this, I employed much of the existing code I had created for the previous graphic. Additionally, I introduced some modifications to add distinctive elements that I deemed necessary, resulting in a new and improved interpretation of the same data set.

plot_parallel_custom <- ggplot(df_long, aes(x = Year, y = n, color = Democracy_Type)) +
  geom_line(aes(group = Democracy_Type), alpha = 0.8, size = 1) +
  scale_x_continuous(limits = c(1789, 2065), breaks = interval_x, minor_breaks = NULL) +
  scale_y_continuous(labels = scales::percent_format(scale = 100), breaks = interval_y) +
  geom_text(
    data = custom_legend,
    aes(x = x + label_distance, 
        y = y + vertical_offset_label, 
        label = label, 
        color = corrected_Democracy_Type),
    size = 2.8,
    family = "Playfair Display ExtraBold",
    hjust = 0) +
  scale_color_manual(values = democracy_colors) +
  labs(
    x = "Years",
    y = "Percentage of Countries",
    title = "Countries that are democracies and autocracies, World.",
    subtitle = "Political regimes based on the classification and assessment by Skaaning et al. (2015)."
  ) +
  theme_minimal() +
  theme(
    panel.grid = element_blank(),
    plot.title = element_text(
      color = "#252525", family = "serif", size = 15, face = "bold", 
      hjust = -0.1, vjust = 0),
    axis.title.x = element_text(
      color = "#252525", family = "serif", size = 12, hjust = 0.44, vjust = -1),   
    axis.title.y = element_text(
      color = "#252525", family = "serif", size = 12,vjust = 3), 
    plot.subtitle = element_text(
      color = "#252525", family = "Lato", size = 9, hjust = -0.1, 
      vjust = 0, face = "italic"),
    axis.text.x = element_text(
      color = "#5b5b5b", size = 9, family = "Playfair Display ExtraBold", 
      vjust = 0 , hjust = c(0, 0, 0, 0, 0.5, 1), margin = margin(t = -9)),
  axis.text.y = element_text(
    color = "#5b5b5b", size = 9, family = "Playfair Display ExtraBold", 
    vjust = 0, hjust = 0, margin = margin(r = -22 )
    )
  ) +
  theme(
    plot.margin = margin(0.4, 0.4, 0.5, 0.7, "cm"), 
  ) +
  guides(color = "none")

plot_parallel_custom

Recognizing the need for enhanced interpretability, I decided to include legends for both the x-axis and y-axis, making it easier to understand the various measures depicted in the plot. To improve readability, I made adjustments such as changing the font color to black. Moreover, I relocated the labels inside the graphic instead of just using segments, believing that this modification would enhance comprehension of each type of democracy and its evolution over time.