About

This is the written version of a YouTube Tutorial created by Spencer Schien. The tutorial recreates a chart found in this NY Times article using R.

Special thanks to Spencer for creating this amazing tutorial! 🙏🙏

To follow similar content from Spencer, subscribe to his YouTube channel.

As a teaser, this is the plot we will be creating:

Libraries

We will use multiple packages from the tidyverse, plus the scales and glue packages. I’ve also included the showtext package, which can optionally be used to load the fonts we will use.

Remember that you can install those packages by running install.packages("package_name") in your R console.

library(tidyverse)
library(scales)
library(glue)
library(showtext) # optional: for custom fonts

Load Packages

Since we are just recreating a chart, we don’t need the underlying data. Every observation is labeled in the original, so we can create our own dataframe of the data we need for plotting.

# Create a vector of our grouping variable.
# This will be the x axis values
household_group <- c(
      "Bottom\n10%", 
      "10th-\n20th", 
      "20th-\n30th", 
      "30th-\n40th", 
      "40th-\n50th", 
      "50th-\n60th", 
      "60th-\n70th", 
      "70th-\n80th", 
      "80th-\n90th", 
      "Top\n10%"
    )

# Turn it into a factor so that it orders correctly.
# Otherwise ggplot will go in alpha order.
household_group = factor(
  household_group,
  levels = household_group
)

# Create a vector for the values, our y variable.
est_change <- c(
  -.039,
  -.012,
  -.004,
  .001,
  .005,
  .008,
  .011,
  .013,
  .015,
  .023
)

# Put it together in a dataframe.
d <- tibble(
  household_group,
  est_change
)

Get the Chart Colors

Now we need to identify the exact colors from the original. For this chart, we can find the colors by inspecting the page and selecting the colored elements to see the assigned hex codes.

I’ve already done this, and the colors are below.

# Colors from the original chart
chart_cols <- c("#d35400", "#58a497")

Get the Font

The NY Times uses its own custom fonts, but we can approximate them using freely available fonts. I have done the legwork to find a font–Libre Franklin, freely available on Google Fonts.

You can either download the font and added it to your system fonts, or you can use the showtext package in R to do so.

# load the font and make it available to ggplot
font_add_google("Libre Franklin")
showtext_auto()

Create the Plot

This chart is a simple bar chart. What might trip some R users up is the very minimal theme and the text labels because this chart employs methods that are far from the ggplot defaults.

Instead of using normal axis labels, we can apply a text geom that serves as those labels. We will also strip away much of the other theme elements to leave the plot looking very crisp.

d |> 
  ggplot(aes(
    household_group, # x axis
    est_change, # y axis
    #  Values < zero use orange fill, values > 0 use green
    fill = ifelse(est_change > 0, chart_cols[2], chart_cols[1]),
    # Same for color, which we will need for text geoms
    color = ifelse(est_change > 0, chart_cols[2], chart_cols[1])
    )) +
  # create the bars
  geom_col() +
  # this hline mimics an x axis
  geom_hline(yintercept = 0) +
  # create the y axis labels above the bars
  geom_text(aes(
    # format labels as percentages
    label = label_percent(.1)(est_change),
    # adjust labels either above or below the bar
    vjust = ifelse(est_change < 0, 1.5, -.5)
    ),
    # change font (changing font in `theme()` won't change it here)
    family = "Libre Franklin", 
    fontface = "bold") +
  geom_text(aes(
    # this will serve as x axis labels
    label = household_group,
    # position them either just below or above x axis
    y = ifelse(est_change > 0, -.005, .005)
    ),
    # adjust font settings
    color = "grey50", 
    lineheight = 1,
    family = "Libre Franklin", 
    size = 3.5) + 
  # this tells ggplot to use the values of the colors we are 
  # assigning above, otherwise it will use defaults
  scale_fill_identity() +
  scale_color_identity() +
  # this keeps ggplot from clipping text that overflows the plot
  coord_cartesian(clip = "off") +
  # remove all theme elements to start
  theme_void() +
  theme( 
    # Set font family, size, color
    text = element_text(family = "Libre Franklin"),
    title = element_text(face = "bold", size = 11),
    plot.subtitle = element_text(face = "plain", color = "grey30"),
    plot.caption = element_text(face = "plain", hjust = 0, size = 10,
                                lineheight = 1.1, color = "grey30",
                                margin = margin(t = 20, b = 5))
        ) +
  # add title, subtitle, and caption
  labs(title = "How the Bill Would Affect Households at Different Income Ranks",
       subtitle = "Estimated annual average change in resources between 2026-34",
       caption = glue::glue(
        "Note: Estimated annual average effect of the House",
        "version of the One Big Beautiful Bill Act on after-tax ",
        "income. Groups are based on income adjusted for household size.  •  ",
        "Source: Congressional Budget Office"
      ) |> str_wrap(110)
    )

Diverging Barplot with R and ggplot2