Creativity, clarity and accessibility with R and ggplot2

data-visualisation
r-how-to

A 3-hour code-along workshop in which we explore new and newish ggplot2 features that allow us to make the most of dataviz to emphasise key patterns in our data stories.

Published

May 6, 2026

Recording

Coming soon!

Slides

View full screen

Code gists and the tricks they include

Full code for the final all singing all dancing graph

The focus is more on showing a combination of {ggplot2} tricks than on dataviz principles - but we covered a lot of that in the recording!

Code 
library(ggplot2)

# Theming and styling elements ----

# Named colour vector, good for everyone's sanity
licorice_gargle_colours <- c(
  "Licorice" = "#39496a",
  "Sugar" = "#c592c7",
  "Highlight" = "#f8f28b",
  "Dark text" = "#1A2231",
  "Light text" = "#383F4C",
  "Background" = "#F8F1F8"
)

# Our theme (we dropped the marquee styling because it didn't play nicely with ggiraph
theme_licorice <- function(base_text_size = 20) {
  theme_minimal(base_size = base_text_size) +
    theme(
      text = element_text(family = "Atkinson Hyperlegible"),
      geom = element_geom(
        ink = licorice_gargle_colours["Light text"],
        borderwidth = 0.5,
        linewidth = 0.2,
        family = "Atkinson Hyperlegible"
      ),
      plot.background = element_rect(
        fill = licorice_gargle_colours["Background"],
        colour = licorice_gargle_colours["Background"]
      ),
      panel.grid = element_line(colour = "white"),
      axis.line.x = element_line(
        colour = licorice_gargle_colours["Light text"],
        linewidth = 0.2
      ),
      axis.text.x = element_text(size = rel(1.2), face = "bold"),
      plot.title = marquee::element_marquee(
        size = rel(1.5),
        colour = licorice_gargle_colours["Dark text"],
        width = 1,
        vjust = 0,
        margin = margin(base_text_size * 1.5, 0, 0, 0),
        style = highlight_style,
        lineheight = 1.05
      ),
      plot.subtitle = ggtext::element_textbox_simple(
        colour = licorice_gargle_colours["Light text"],
        lineheight = 1.3,
        margin = margin(base_text_size * 0.75, 0, base_text_size, 0)
      ),
      legend.position = "none",
      plot.margin = margin_auto(base_text_size * 1.5)
    )
}

# Our background-to-red gradient
bg_to_red <- grid::linearGradient(
  colours = c(licorice_gargle_colours["Background"], "red"),
  x1 = 0,
  y1 = 0,
  x2 = 0,
  y2 = 1,
  group = TRUE
)

# Marquee styling (we don't end up using this)
highlight_style <- marquee::classic_style(weight = "bold") |>
  marquee::modify_style(
    "sugar",
    background = licorice_gargle_colours["Sugar"],
    # US spelling of colour/color required!
    color = licorice_gargle_colours["Dark text"],
    padding = marquee::trbl(marquee::em(0.1))
  ) |>
  marquee::modify_style(
    "licorice",
    background = licorice_gargle_colours["Licorice"],
    color = "white",
    padding = marquee::trbl(marquee::em(0.1))
  )

# Getting the data ready ----
# For our first plot
tidied_data <- medicaldata::licorice_gargle |>
  dplyr::mutate(
    gender = factor(
      dplyr::case_when(
        preOp_gender == 0 ~ "Male",
        preOp_gender == 1 ~ "Female"
      )
    ),
    id = dplyr::row_number(),
    treatment = factor(dplyr::case_when(
      treat == 0 ~ "Sugar",
      treat == 1 ~ "Licorice"
    ))
  ) |>
  dplyr::left_join(
    babynames::babynames |>
      dplyr::group_by(sex) |>
      dplyr::mutate(id = dplyr::row_number()) |>
      dplyr::select(sex, name, id) |>
      dplyr::mutate(
        sex = factor(dplyr::case_when(
          sex == "F" ~ "Female",
          sex == "M" ~ "Male"
        ))
      ),
    by = dplyr::join_by(gender == sex, id == id)
  ) |>
  dplyr::rowwise() |>
  dplyr::mutate(
    praise = paste(
      sample(
        praise::praise_parts$adverb_manner,
        1
      ),
      sample(praise::praise_parts$adjective, 1)
    )
  ) |>
  dplyr::ungroup()

# For our timeline plot
timeline_data <- tidied_data |>
  dplyr::select(
    gender,
    name,
    praise,
    treatment,
    preOp_age,
    pacu30min_throatPain,
    pacu90min_throatPain,
    postOp4hour_throatPain,
    pod1am_throatPain
  ) |>
  tidyr::pivot_longer(
    -c(gender, name, praise, treatment, preOp_age),
    names_to = "timeline",
    values_to = "pain"
  ) |>
  dplyr::mutate(
    timeline = factor(
      timeline,
      levels = c(
        "pacu30min_throatPain",
        "pacu90min_throatPain",
        "postOp4hour_throatPain",
        "pod1am_throatPain"
      ),
      labels = c("30mins", "90mins", "4hours", "Next morning"),
      ordered = TRUE
    )
  )

# For our rectangles in the timeline data plot
mean_pain <- timeline_data |>
  dplyr::group_by(timeline, treatment, gender) |>
  dplyr::summarise(mean_pain = mean(pain, na.rm = TRUE)) |>
  dplyr::ungroup()


# The plots ----
interactive_cough_plot <- ggplot(
  tidied_data,
  aes(x = gender, y = pod1am_cough, fill = treatment)
) +
  geom_rect(
    data = tidied_data |> tail(1),
    aes(xmin = I(0.1), xmax = I(0.9), ymin = 0, ymax = 3.5),
    fill = bg_to_red,
    alpha = 0.1
  ) +
  ggiraph::geom_point_interactive(
    aes(
      tooltip = paste0("<b>", name, "</b>, ", preOp_age, ", is ", praise),
      data_id = name,
      group = treatment
    ),
    alpha = 0.8,
    position = position_jitterdodge(
      jitter.height = 0.05,
      jitter.width = 0.2,
      dodge.width = 0.25
    ),
    shape = 21,
    size = 5
  ) +
  stat_summary(
    aes(ymin = after_stat(y), ymax = after_stat(y), colour = treatment),
    fun = function(x) mean(x, na.rm = TRUE),
    geom = "crossbar",
    width = 0.5
  ) +
  stat_summary(
    aes(
      label = janitor::round_half_up(after_stat(y), 2),
      group = treatment
    ),
    fun = function(x) mean(x, na.rm = TRUE),
    geom = "text",
    position = position_nudge(y = 0.06),
    colour = "black",
    size = 16,
    size.unit = "pt"
  ) +
  geom_text(
    data = tibble::tibble(
      y_coord = c(0, 1, 2, 3),
      severity = c("No cough", "Mild", "Moderate", "Severe cough")
    ),
    aes(
      x = I(0.5),
      y = y_coord,
      label = severity,
      fill = NULL
    )
  ) +
  labs(
    title = paste0(
      "Patients who gargled <span style='color:",
      licorice_gargle_colours["Licorice"],
      "'>Licorice</span> reported a slightly less severe cough than those who gargled <span style='color:",
      licorice_gargle_colours["Sugar"],
      "'>Sugar</span>"
    ),
    subtitle = "The effect was similar across Female and Male patients, on the morning after surgery."
  ) +
  scale_fill_manual(values = licorice_gargle_colours) +
  scale_colour_manual(values = licorice_gargle_colours) +
  theme_licorice() +
  theme(
    plot.title = ggtext::element_textbox_simple(
      size = rel(1.5),
      face = "bold",
      margin = margin(30, 0, 15, 0),
      vjust = 0
    ),
    axis.title.y = element_blank(),
    axis.text.y = element_blank(),
    axis.title.x = element_blank(),
    legend.title = element_blank(),
    panel.grid.major.x = element_blank(),
    panel.grid.major.y = element_blank()
  )

# This is used by layout = "fixed" later
update_geom_defaults(
  geom = "point",
  aes(
    colour = "white",
    fill = licorice_gargle_colours["Light text"],
    shape = 21,
    size = 4,
    alpha = 0.5
  )
)

timeline_graph <- ggplot(timeline_data, aes(x = timeline, y = pain)) +
  geom_rect(
    data = mean_pain,
    aes(
      xmin = as.numeric(timeline) - 0.5,
      xmax = as.numeric(timeline) + 0.5,
      ymin = -Inf,
      ymax = Inf,
      fill = "red",
      alpha = mean_pain
    ),
    inherit.aes = FALSE
  ) +
  scale_alpha(range = c(0.05, 0.25)) +
  # Slight fudge!
  geom_point(
    layout = "fixed",
    alpha = 0.1
  ) +
  # Move lines to back for easier interaction
  ggiraph::geom_line_interactive(
    aes(
      group = name,
      colour = treatment,
      data_id = name,
      tooltip = paste0("<b>", name, "</b>, ", preOp_age, ", is ", praise)
    ),
    alpha = 0.3
  ) +
  ggiraph::geom_jitter_interactive(
    aes(
      x = timeline,
      y = pain,
      colour = treatment,
      data_id = name,
      tooltip = paste0("<b>", name, "</b>, ", preOp_age, ", is ", praise)
    ),
    height = 0.1,
    width = 0.1,
    alpha = 0.6
  ) +
  scale_colour_manual(values = c("Sugar" = "#c592c7", "Licorice" = "#39496a")) +
  scale_x_discrete(labels = function(x) {
    # Thank you, Claude!
    stringr::str_wrap(gsub("([0-9])([a-zA-Z])", "\\1 \\2", x), 5)
  }) +
  facet_grid(treatment ~ gender) +
  labs(
    title = "The Licorice effect on pain reduction was most visible within the first 4 hours after surgery",
    subtitle = "Pain was rated 0 = no pain to 10 = worst pain, but the maximum pain reported was 6."
  ) +
  theme_licorice() +
  theme(
    legend.position = "none",
    axis.title.y = element_blank(),
    axis.title.x = element_blank(),
    strip.text.y = element_text(angle = 0),
    # To get the two titles to look the same; I probably should have edited our reusable theme instead!
    plot.title = ggtext::element_textbox_simple(
      size = rel(1.5),
      face = "bold",
      margin = margin(30, 0, 15, 0),
      vjust = 0
    )
  )

# Tie the plots together ----
# The key here is a common data_id. If you look carefully, you'll see
# we have both a female and a male patient called Ora.
# Be careful what you choose as your data_id!

library(patchwork)

ggiraph::girafe(
  ggobj = interactive_cough_plot +
    timeline_graph +
    plot_layout(heights = c(1, 1), widths = c(1, 1), nrow = 1),
  height_svg = 15,
  width_svg = 25,
  options = list(
    ggiraph::opts_hover(
      css = "stroke:#f8f28b;stroke-width:5px;stroke-opacity:1;opacity:1;fill:#f8f28b;"
    ),
    ggiraph::opts_tooltip(
      css = "background-color:#383F4C;color:#F8F1F8;padding:7.5px;letter-spacing:0.025em;line-height:1.3;border-radius:5px;font-family:'Atkinson Hyperlegible';",
      opacity = 0.9,
      offy = 10
    ),
    ggiraph::opts_sizing(width = 1),
    ggiraph::opts_hover_inv(css = "opacity:0.5")
  )
)

Reuse

Citation

For attribution, please cite this work as:
“Creativity, Clarity and Accessibility with R and Ggplot2.” 2026. May 6, 2026. https://www.cararthompson.com/talks/rmedicine2026-ggplot2-creativity-clarity-accessibility/.